Recursive Neural Network-Based Market Demand Forecasting Algorithm for Calligraphy Practice Products

Yi Xue

doi:10.1155/2022/8086789

Recursive Neural Network-Based Market Demand Forecasting Algorithm for Calligraphy Practice Products

Xue, Yi 2022-01-05 00:00:00 Hindawi Journal of Mathematics Volume 2022, Article ID 8086789, 10 pages https://doi.org/10.1155/2022/8086789 Research Article Recursive Neural Network-Based Market Demand Forecasting Algorithm for Calligraphy Practice Products Yi Xue Jilin University of Architecture and Technology, Jilin, Changchun 130000, China Correspondence should be addressed to Yi Xue; 2016121641@jou.edu.cn Received 17 November 2021; Revised 30 November 2021; Accepted 6 December 2021; Published 5 January 2022 Academic Editor: Naeem Jan Copyright © 2022 Yi Xue. *is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Intoday’ssociety,calligraphy,whichreﬂectsone’sbasicwritingskills,isbecomingmoreandmoreimportanttopeople.Peopleare inﬂuenced by calligraphy in their studies, work, etc. Improving calligraphy writing skills has become one of the key directions for developing one’s abilities at this stage. As an important means of improving writing skills, calligraphy practice products are attractingmoreandmoreattentionandpurchases.Inparticular,inrecentyears,asthemarketeconomyhasdevelopedinadeeper direction, people’s demand for calligraphy practice products has diversiﬁed and calligraphy practice product companies have launched a varietyof productsto meetthe public’scalligraphypracticeneeds in ordertoadapt tothe reality of consumer demand. However, with the development of the Internet culture industry and inﬂuenced by objective factors such as school holidays and seasons, the current market demand for calligraphy practice products is rapidly and dynamically changing, making market changes diﬃcult to grasp and leading to poor sales, which directly aﬀects the proﬁts of calligraphy practice product-related companies. *e artiﬁcial intelligence neural network method realizes the nonlinear relationship between the input and output of sampledatathroughtheself-learningabilityofeachneuronandhasacertainnonlinearmappingabilityinprediction,whichplays a great role in the market demand prediction of many commercial products. Based on this, this paper proposes a recursive neural network-based algorithm topredict the future demandand development trend of calligraphy practice products through extensive and in-depth research, so as to provide positive and beneﬁcial guidance for enterprises’ future production and sales. vast sea of Chinese culture. *e art of calligraphy has always 1. Introduction been respected by the nation as an elegant, erudite, and China has been a cultural power since ancient times, and nurturing art form. *e art of Chinese calligraphy is a after more than a thousand years of historical evolution, the manifestation of Chinese culture and is one with the spirit of major ancient Chinese civilizations have long inﬂuenced the Chinese nation. *e spirit of Chinese culture is the unity each other and merged. Today, a China with a splendid of heaven and man, the valuing of harmony, and the culture stands in the East of the world with its colorful shunningofcenter.*eChinese culture’svalueof“harmony is precious” is perfectly embodied through the beauty of cultural elements. An important part of Chinese culture is writing, and the diﬀerent aesthetic forms of writing form the harmony in the art of calligraphy [2]. Calligraphy practice art of calligraphy [1]. *e art of Chinese calligraphy, with its products collection is shown in Figure 1. long and unbroken history, has attracted generations of Chinese calligraphy, with its ability to express the calligraphers to cultivate and tread the paths of the ancients, richness and complexity of human thought and emotion in to experience the essence of the culture that has been the simple shape of lines, has great aesthetic value. *e study precipitated for thousands of years, and on the other hand, of Chinese calligraphy is a great help in understanding to read ten thousand books, travel ten thousand miles, Chinese art and culture. Calligraphy exists alongside the capture the world, cultivate their bodies, and express it in written word and is concerned with brushwork, strokes, and theirbrushstrokes,creatingastyleofcalligraphyappropriate strokes of meaning. With the help of Chinese characters, to the times. Together with other forms of art, they form a calligraphy is an expression of sentiment and sentimentality 2 Journal of Mathematics Figure 1: Calligraphy practice products collection. and has a strong national character and high artistic taste. cultivate the body and nourish the mind. Calligraphy is indeed a great source of health beneﬁts, allowing the whole Chinese calligraphy has been steeped in thousands of years and is ubiquitous in Chinese life. It is closely related to bodytomove,thewaistandarmstrengthandeyestrengthto be harmonized, and the rigidity and ﬂexibility to be com- residential culture, landscape culture, educational culture, religious culture, political culture, coin culture, folk ritual bined with movement and stillness, as well as strengthening customs, and the arts of painting, architecture, dance, arts, the body and brain, eliminating fatigue, regulating the and crafts [3]. As shown in Figure 1, Chinese calligraphy is a psyche,andquietingthemind.Calligraphersandcalligraphy unique visual art that still exudes a fascinating artistic charm education work have also become a highly respected pro- after thousands of years. *e artistic beauty of Chinese fession in society. *e Chinese Calligraphers Association, calligraphy is composed of two elements: form and spirit. which consists of national calligraphers, seal carvers, cal- *e beautiful and distinct artistic eﬀects and infectious ligraphy theorists, calligraphy educators, and calligraphy power of calligraphy reﬂect a unique national style. *e activity organizers and management workers, has become a national professional organization and a group member of composition of traditional painting and calligraphy requires the author not only to make a subjective and rational the Chinese Federation of Literary and Art Circles. Callig- judgment but also to take the various elements of the picture raphy enthusiasts have relatively good social recognition. and to achieve a dialectical unity between the whole and the *is intangible social recognition and support provide great parts so that the picture can achieve balance and harmony, assistance in the spread, learning, and development of which can be regarded as the earliest theory of composition calligraphy [5]. in Chinese painting [4]. *e aesthetic value of the art of Chinese calligraphy has always been an excellent tra- calligraphy has many applications in home decoration, ditionalculturethatweareproudof,withthousandsofyears where its content has a subtle eﬀect on the human spirit, of history. As life has improved, many parentshave begun to gaining knowledge and culture, drawing inspiration and pay attention to the cultural development of their children. strength. At the same time, calligraphy with its unique Learning calligraphy can lay the foundation for children’s writing and can improve their overall quality. China’s means of art life and service life is widely used in the design of patterns for product packaging. *e art of Chinese cal- contemporary design started late, and so did the stationery ligraphy,whichonthesurfaceappearstobeverycommon,is industry’sdesignsystem,withalargeproportionofdomestic nothing more than taking a brush and dipping it in ink to stationery manufacturers borrowing or directly copying write Chinese characters on paper (of course, other writing foreign stationery products in the early days. However, as a carriers are sometimes used), butit is this extreme simplicity local cultural product, it is diﬃcult to ﬁnd a mature product and richness of variation that has made calligraphy less from abroad that is worth learning from. With the rapid attractive in contemporary times due to the advancement of development of the Internet culture industry, calligraphy modern science and technology, especially the spread of practice products are now combined with modern design to computers and the reduction of handwriting. It has become create new developments that shine in a diﬀerent light [6]. one of the most popular art forms in contemporary China, As a representative of traditional culture, the calligraphy practice product industry is also facing unprecedented with the largest number of participants and the widest audience. Calligraphy is widely recognized for its ability to impact, challenges, and opportunities in the Internet era, Journal of Mathematics 3 and it is worthwhile to deeply explore the artistic value of First, consumables such as ink, paper, and inkstone are calligraphy and promote it, thus creating economic value relatively inexpensive compared to other hobbies, such as and prosperity for the calligraphy industry. In recent years, musical instruments and photography equipment. *e vast with the rapid development of computer technology, the majority of consumers can aﬀord them. Beginners can Internet is deeply integrated with various traditional in- basically get all the materials they need to learn calligraphy dustries. *e Internet economy,represented by the booming for under $1,000. With the rise of productivity and the e-commerce, has become an important engine of economic economy, the cost of consumables for calligraphy practice is development. *is will provide great development potential no longer an issue that hinders the spread of calligraphy and broader development space for the rise of the Internet learning. Secondly, the calligraphy training industry is also economy. *e Internet has changed people’s lifestyles and in full swing with economic development. As more and brought more possibilities for consumption upgrading and more families reach the well-oﬀ level and have a relatively industrial development. *e further integration of the In- goodincome,parentshavethefundsandabilitytosendtheir ternet with traditional industries has changed the way the children to calligraphy training courses to learn calligraphy. *e cost of calligraphy education is in the middle of the original industries produce,sell, andcommunicate, bringing new opportunities for development. As a typical represen- range of hobby learning costs and is becoming more and tative of traditional industries, the calligraphy practice more acceptable to parents and families. *e good economic product industry is also integrating Internet technology, development and the increase in consumption power have quietly undergoing the biggest change in thousands of years. created a good development environment and opportunity Artiﬁcial intelligence is a new technical science that studies for the calligraphy training industry. In addition, economic and develops theories, methods, technologies, and appli- development has led to the development of the Internet and cation systems used to simulate, extend, and expand human online education, and payment for knowledge has become intelligence. Artiﬁcial intelligence-based market demand theconsensusofthetimes.Onlineeducationandonlinepaid forecasting method refers to the method of using artiﬁcial courses in calligraphy are also being accepted by more and intelligence techniques, mainly machine learning and deep more adult learners, all of which have brought great pros- perity and opportunities to the calligraphy practice products learning, etc., to simulate industry experts for market de- mand forecasting, which diﬀers from fundamental analysis market. Practitioners of calligraphy practice products have and technical analysis methods in terms of modeling also enjoyed the dividends of economic development, and methods, and focuses on the use of artiﬁcial intelligence calligraphy, an important art form, has gradually become a techniques to uncover potential connections between data leisurely and elegant way of life for the general public after from various sources in the market and market movements. the majority of people have solved the problem of food and In order to adapt to the new environment of the Internet clothing [7]. culture industry, it is necessary to design and study new Withthedevelopmentofsocietyandtheadvancementof intelligent ﬁeld demand forecasting algorithms for the cal- technology and culture, the calligraphy industry has also ligraphy practice product market using artiﬁcial intelligence seen unprecedented growth. Various calligraphy practice techniques. products have become abundant, the four treasures of In this paper, we start by presenting related works in calligraphy supplies continue to innovate, and new teaching Section 2; then, in Section 3, we discuss an artiﬁcial intel- models and online classes have become popular with the ligence neural network method that has certain nonlinear advancementofvideoandlivestreamingtechnology.Online mapping ability in prediction; and for the concerned shopping, online payments, and interpersonal trust have method,wedoexperiments inSection4andevaluationsand taken calligraphy practice products from oﬄine to online. It ﬁnally put a conclusion in Section 5. can be said that the calligraphy industry has also enjoyed the great convenience of the Internet and technological ad- vances, and the dissemination of calligraphy has become 2. Related Work easier and more convenient. However, the calligraphy 2.1. Calligraphy Practice Products Market Analysis. As an market still has its limitations [8]. Calligraphy is a niche important part of art and culture, calligraphy has gradually industry with a small group of learners, and although cal- ligraphy can be found everywhere in life, it is still a minority moved from the study to the marketplace, forming the basic chain of industrialization, the emergence of specialized of people who actually learn and come into contact with calligraphy, so the small consumer base and low degree of calligraphic art dealers and collectors, and the involvement of merchants in the operation of calligraphic artwork that consumer demand are the biggest obstacles to building a calligraphy industry chain. Compared to the entertainment has to some extent further guided the development of calligraphy in the direction of industrialization. At the same and leisure industries, such as movies, music, dance, and time, the development of a related industrial chain plus the food, which have a high degree of popularity, calligraphy is framing of calligraphic works and the trading of the Four still relatively inactive. Calligraphy is a quiet art, limited to Treasures of calligraphy market gradually took shape. As the the line and visual appreciation, lacking auditory and gus- tatory stimulation, with limited impact and limited inﬂu- national economy is developing rapidly and people’s con- sumption levels are increasing, the art of calligraphy is also ence, and with the popularity of computer fonts, the practicality of calligraphy is even weaker. Many people are seeing better opportunities for development. People have more time and money for calligraphy-related consumption. not involved, making market activity and demand 4 Journal of Mathematics servants, etc.). K12 calligraphy post is the largest segment of lukewarm. *e cycle of production (creation), sales, and consumption of calligraphy are slow, circulation is tiny, age the entire calligraphy post industry, accounting for more than 60% of the market. At present, China’s character post groups are large, and innovation is slow. Calligraphy products are also nondepleting; many calligraphy products industry is mainly led by several well-known calligraphers, can be used for years, and many calligraphy paintings can be and there are mainly two types of calligraphy practice preserved for years, which makes the replacement of new character post usage forms, one is the tracing character post, calligraphy products slow, and low consumption and low and the other is the water writing character post. *eir attrition make it diﬃcult for demand for calligraphy common drawback is the lack of interaction, the use of a productstorise.Calligraphyisanelegant,high-end,spiritual single mode, the use of boring, and not being able to write a variety of fonts. It is based on thelack of interaction between art with a high skill threshold, requiring a certain level of practice, cultural heritage, and knowledge to resonate, love, the existing handwriting posters and the practitioner where the laser projection technology and VBK keyboard tech- and appreciate. Calligraphy also has regional and cultural limitations, limited to a few countries such as China and nology are integrated into the interactive laser projection calligraphy practice product, which has the advantages of Japan, and from an international perspective, it lacks global demand and circulation. All of this has hindered the rapid easy writing, strong interactivity, and assistance in cor- development of the calligraphy market. *e calligraphy recting posture. It has three practice modes. First, learning industry is a more stable, long-term lukewarm state of mode: the laser emitter projects the trajectory and strokes of development, with an overall stable Baidu search index and calligraphy on the paper, visually demonstrating the writing information index with a slight increase. Calligraphy is a technique; second, copying mode: the laser emitter projects traditional culture of thousands of years, and even with the the outer outline of the font on the paper, and the user can directly copy the practice; third, practice mode: the laser impact of the Internet and artiﬁcial intelligence, it will continuetobe relativelystableandwillnotexperiencemajor emitterprojectsthemetergridonthepaper,andtheusercan practice writing, and after the writing is completed, the ups and downs [9]. *e calligraphy industry is inactive, and the calligraphy auction industry is relatively sluggish, with product’s intelligent scoring function will rate and voice correct the user’s writing. After writing, the product’s in- fewer inﬂuential calligraphy works and calligraphy masters, although precious calligraphy works can fetch high prices. telligent scoring function will rate the user’s writing and *ecalligraphyindustryisindisarray,withsmallcalligraphy voicecorrection.Atthesametime,thesystemissetupwitha practice product manufacturers stealing and evading taxes, library of diﬀerent types of fonts, so users can choose dif- andcalligrapherswithlowtaxcapacityorconductingprivate ferent types of fonts for practice. *e interactive laser transactions without paying taxes. Numerous small and projection calligraphy practice product makes up for the scattered calligraphy training institutions are popping up, shortcomings of the existing calligraphy posters while enriching the functions of the calligraphy practice tool to and paying taxes is not regulated enough. *e state does not havemuchrevenuetospendonthecalligraphyindustry,and better meet the needs of consumers. In terms of product functionality, the interactive laser projection calligraphy there is no major investment. In the calligraphy industry analysis, the competition in the same industry for sales of practice product has a great advantage in that it allows for calligraphy products also needs to be mastered. Oﬄine good human-computer interaction and can meet the re- competition can be judged by market research; whether quirements of users with diﬀerent fonts, while being able to there are local calligraphy training institutions, calligraphy detectandcorrectthewordswrittenbytheuser,preservinga galleries, and calligraphy stores, etc., and how the customer realistic writing experience. In terms of functionality, this ﬂow and sales are can be investigated. Competition on product is the equivalent of a calligraphy postcard with online e-commerce platforms can be obtained through big thousands of fonts in one and a calligraphy teacher at the data analysis tools on e-commerce platforms, allowing for a same time, providing a one-stop service for learning, practicing, and correcting calligraphy, which is not possible clear, accurate, real-time grasp of competition in the same industry. *e Internet and big data provide tremendous and with the calligraphy practice products currently available on the market. *e interactive projection calligraphy practice convenient help for marketing analysis and business deci- sions. In conclusion, calligraphy practice products have product has advanced technical support to ensure that its enjoyed the dividends of the Internet technology era, but function is achieved. *e product uses the currently mature there are also great limitations and space, which requires the laser projection technology and VBK keyboard technology, joint eﬀorts of every calligraphy lover and calligraphy in- with a built-in laser emitter to project the interface infor- dustry practitioners to pioneer and innovate and constantly mation on the surface of any paper and then track the ﬁnger explore to make it prosperous and strong. movements through infrared technology to ﬁnally complete theacquisition of inputinformation. High-tech productsare Calligraphy practice is often easy to start and diﬃcult to become a master, or it is easy for beginners to practice a trend for future product development, and consumer enthusiasm for high-tech products is always growing. When calligraphy through text posters. China’s calligraphy post industry according to user classiﬁcation is mainly divided high-tech elements are incorporated into a product, then the product can stand out among similar products. *e inter- into the K12 calligraphy post market (the main users are kindergarten to high school), adult calligraphy post market, active laser projection calligraphy practice product is pre- and professional calligraphy post market (the main users ciselytheintegrationoflaserprojectiontechnologyandVBK with certain career characteristics, such as doctors, civil keyboard technology that makes the product stand out Journal of Mathematics 5 that characterize trends. In daily applications, RNN models among calligraphy practice products. With the increasing popularity of high-techproducts, interactive laser projection are widely used in tasks such as speech recognition and machine translation, bringing convenience to people’s lives. calligraphy practice products will be favored by many consumers. According to the “Guideline for Primary and In the biomedical ﬁeld, sequence data are even more diverse Secondary School Calligraphy Education” published on the and large in scale, in which RNNs are widely used for tasks website of the Ministry of Education, elementary school such as health monitoring, auxiliary diagnosis, and infor- grades 3 to 6 are required to schedule one class period per mation extraction, making great contributions to medical weekforbrushwritingstudy.Accordingtostatisticsreleased development as well as human health. Most biomedical by the National Bureau of Statistics on February 24, 2015, it signals are time-varying (time-variant) nonlinear signals, can be seen that 93.6 million students were enrolled in including electrocardiogram (ECG), electroencephalogram generalelementaryschoolsnationwidein2015.*ereissuch (EEG), and electromyogram (EMG). Based on ECG signals, RNN can be used for arrhythmia diagnosis. Based on EEG a huge group of calligraphy practitioners among children. *e latest survey results released by Regus, a world-re- signals, RNN can be used for sleep signal classiﬁcation. In addition, in the brain-computer interface, RNN can be used nownedoﬃcesolutionsprovider,saidthatthestresssuﬀered by oﬃce workers in mainland China in the past year ranked to decode EEG signals and improve the grasping accuracy of ﬁrst in the world. Nowadays, oﬃce workers’ life is neural prostheses. In recent years, the multimodal (multi- accelerated and their work life is irritating; they are also modal) biomedical signal utilization approach has become a eager to seek a new way to reduce stress. Practicing cal- trend. For example, RNN can simultaneously utilize EEG ligraphy can cultivate the mind, practice qi, and beneﬁt the signals as well as functional near-infrared spectroscopy (f brain,nurturerespect,andgreatlyreducethestressfromlife. NIRS) for epilepsy monitoring. Text in biomedicine is even more complex and diverse, and it is natural that RNNs can With the development of society and the aging of the population, more and more elderly people are experiencing perform natural language processing (NLP) on such data [12]. For example, RNN-based biomedical named entity “retirement syndrome” because they do not know how to organize their lives after retirement. In addition to con- recognition (BNER) technology can help researchers extract usefulinformationfrommassivebiomedicaltexts.Inclinical sideringretirementasarest,theelderlyshouldalsomaintain a certain level of tension so that life does not become boring. practice, medical images including ultrasound images, CT *is shows that there is a huge demand for calligraphy images, and MRI images are an important basis for disease practice tools. In this regard, we did an interview survey, by diagnosis. Based on ultrasound image sequences, RNN interviewing some parents, teachers, students, oﬃce models can signiﬁcantly improve the detection accuracy of workers, and retired elderly people, the survey results show prostate cancer. In computer-aided detection (CAD), RNN thatprimaryand secondaryschoolstudentshavetheneed to can improve the accuracy of early breast cancer detection using current X-ray scans as well as previous scans. At the practice calligraphy, and it is very important for the de- velopment of students. Working people have to sign con- microscopic scale, RNNs have also been successful in the task of target tracking in time-lapse cell image sequences. In tracts, agreements, examinations for titles, etc., and they need to improve their writing skills urgently, and the de- the clinical setting, electronic health record (EHR) data mand for word practice is also great. Secondly, retired se- containssequencesofdiﬀerentmodalities,whichistypicalof niors: surveys show that many retired seniors prefer to read multimodal data. Based on EHR, RNN can perform disease and write then to play cards and watch TV and other similar prediction and also predict the risk of readmission of dis- activities. So according to the survey results, the target charged patients over a period of time, providing the pos- market for calligraphy practice products should be basically sibility of early initiation of targeted intervention programs located in primary and secondary school students, oﬃce for patients at risk of readmission, eﬀectively reducing the workers, and retired seniors [10]. probability of readmission, and controlling the cost of care. In recent years, dynamic treatment recommendation sys- tems based on large-scale electronic health records have emerged as the key to the successful improvement of actual 2.2. Recursive Neural Networks. Artiﬁcial neural network (ANN) has been named deep learning at the time of rapid clinicaloutcomes.However, thesesuccessesdonotconceala development of computer storage, communication, and series of problems that arise during the backpropagation of computing power and has set oﬀ the third artiﬁcial intel- error signals in the depth structure when training RNN ligence boom in the world’s history with unprecedented models with gradient descent, mainly including the van- scale. “Deep” gives neural networks the ability to abstract ishing and exploding gradient problem and the gradient information in a hierarchical way, giving them the “intui- conﬂict. *e former is the most common problem in the tion” that people have [11]. Overnight, many ﬁelds that had general feed-forward neural system. *e former is also beenﬁrmlydominatedbyﬁne-grainedmathematicalmodels frequently mentioned in general feed-forward neural net- work (FNN) research [13]. Before the popularization of let go of their previous obsession with equations and their prejudice against the inscrutable black box and jumped into Rectiﬁed Linear Unit (ReLU), FNNs with deeper structures were often diﬃcult to train due to these problems (especially the deep learning bandwagon. In this transformation, a recursive neural network (RNN) is responsible for modeling the gradient vanishing problem). Unlike FNNs, the depth sequential data. Common serial data include signals, text, structure of RNNs is inherent, because even an RNN model video image sequences, and serial data such as stock prices that contains only a single layer of hidden units in space 6 Journal of Mathematics becomesverydeepwhenitisexpandedintime.RNNmodels Forget Gate have the ability to remember, are turing complete, and can Self-recurrent theoretically learn dependencies at any time interval in a connection sequence. However, the long-time memory capability of RNNs is not reliable due to the aforementioned problems. Memory cell Memory cell Speciﬁcally, it is very diﬃcult to rely on gradient descent to Input output cell train RNN models to learn long-time dependencies, and early RNN models with simple structures are often deep in the quagmire of gradient disappearance. *e Long Short- Input Gate Term Memory (LSTM) proposed at the end of the last Output Gate century largely alleviated the gradient disappearance Figure 2: Internal structures of neurons in recurrent neural problem, which made RNNs shine in this century’s boom, networks. and the LSTM structure itself became the standard con- ﬁguration of most RNN models. However, LSTM has not fundamentally changed the dilemma that the RNN family is created to address the need for sequential inputs. Long in, and gradients still vanish in the face of longer time series. Short-Term Memory (LSTM) is essentially a speciﬁc form of On the other hand, the gradient explosion problem has recursive neural network (RNN). Inaddition to containing a neverbeenproperlysolved.Atthetimeofwritingthispaper, memory module, LSTM solves the RNN long-term de- the third wave of neural networks is still going on. *e pendency problem by adding a threshold (Gated RNN) to amount of data available for learning by various neural solve the problem that the LSTM adds input gate, output models is increasing dramatically, and new sequence data gate, andforget gate tothe basic structure of RNN [15]. Each and their corresponding learning tasks are emerging. *e of these three logic control units is connected to a multi- long-time dependencies embedded in the sequence data are plication element (see Figure 2), and the input and output of becoming richer, longer in time span, and more obscure in the information ﬂow and the state of the memory cell are presentation, and capturing these long-time dependencies controlled by setting the weights at the edges where the eﬃciently and accurately is the key to dealing with such memory cell of the neural network is connected to the other problems. On the other hand, neural network models are parts. *e speciﬁc structure is shown in Figure 2. increasing day by day. Increasing the network size by To note, the input gate denoted as i controls whether stacking model layers and increasing link lengths usually information ﬂows into the memory cell whereas the forget acceleratesthedecline ofmodelmemory.Inthiscontext and gate controls whether the information in the memory cell at historical trend, it is urgent to investigate the RNN archi- the previous moment accumulates into the memory cell at tecture and related algorithms with long-time memory ca- the current moment, and it is denoted asf . *e output gate pability. *e method proposed in this paper is the market denoted as o controls whether the information in the demand prediction of calligraphy practice products with memory cell at the current moment ﬂows into the currently long- and short-term memory network architecture. hidden state h . We recall also the cell “memory cell”, which represents the memory of the neuron state and makes the 3. Method LSTM cellhavetheabilitytosave,read,reset, andupdatethe long history information, denoted as c . At moment t, the *e Internal structure of neurons in recurrent neural net- LSTM neural network is deﬁned by the following equation: works is shown in Figure 2. f � sigmoid􏼐W · 􏼂h , x 􏼃 + b 􏼑, A neural network is a model with strong learning ability, t f t−1 t f especially eﬀective in dealing with pattern recognition, in- i � sigmoid W · 􏼂h , x 􏼃 + b 􏼁, t i t−1 t i telligent control, and other problems. In recent years, with o � sigmoid W · h , x + b , the booming development of computer science and tech- 􏼂 􏼃 􏼁 t o t−1 t o (1) nology and hardware devices, neural networks have grad- 􏽥 c � tanh W · 􏼂h , x 􏼃 + b 􏼁, t c t−1 t c ually become a hot research topic in the ﬁeld of artiﬁcial c � f ∗ c + i ∗ c , intelligence [14]. However, Full Connected Neural Network t t t−1 t t (FCNN) has the limitation that the nodes of the same layer h � o ∗ tanh c􏼁 . t t t are not connected to each other, and FCNN cannot be used when the information of the previous moments of the se- *e recursive neural network timing diagram is shown quence is needed. Since FCNNs cannot share features be- in Figure 3. tween diﬀerent locations of a sequence, they can only handle In addition to the previously mentioned i f , o , and c , t t t t one input individually, i.e., there is no relationship between W represent the recursive connection weights of their the previous input and the next input, and they cannot corresponding gates, and sigmoid and tanh are the two handle inputs that are related back and forth in time or activation functions. In the training process of the LSTM space. However, many learning tasks need to deal with neural network, ﬁrstly, the data features at moment t are sequential information, such as time series prediction and input to the input layer and the results are output through task-based dialogues, which require the model to learn from the excitation function. *e output results, the output of the sequential inputs. Recursive Neural Network (RNN) was hidden layer at moment t−1, and the information stored in Journal of Mathematics 7 C C C C C t-n t-n+1 t t+1 t+2 c c c c h h h h h t-n t-n+1 t t+1 t+2 h h h h x x x x t-n t-n+1 t t+1 Figure 3: Recursive neural network timing diagram. the cell unit at moment t−1 are input into the nodes of the z � σ V · s , (2) t t LSTM structure, and the data are output to the next hidden layer or output layer through the processing of input gate, s � δ U · x + W · s , (3) t t t−1 output gate, forget gate, and cell units, the results of the nodesoftheLSTMstructureareoutputtotheneuronsofthe where V is the weight matrix of the output layer; σ is the output layer, and the calculation of backpropagation error output layer activation function; U is the weight matrix of updates each weight. *e overall structure is shown in the input x; W is the weight matrix of the hidden layer state Figure 3. s at moment t−1 as the input at moment t; g is the hidden t−1 A simple recursive neural network consists of an input layeractivationfunction.Loopingequation(3)intoequation layer, a hidden layer, and an output layer. For a given input (2), we get sequence x � [X , X , . . . , X ], at moment t, the hidden 1 2 t layer state is s and the output value is z , as follows: t t z � V∗δ U∗ x + W∗δ U∗ x + W∗δ U∗ x + W∗δ U∗ x + · · ·􏼁􏼁􏼁􏼁. (4) t t t−1 t−2 t−3 *e LSTM diﬀers from the RNN in that it adds a one layer up; (3) calculating the gradient of each weight “processor” to the algorithm to determine whether the in- according to the corresponding error term. formation is useful or not. A cell has three gates, called input gate,forgetgate,andoutputgate.Whenamessageentersthe 4. Experimentation and Evaluation LSTMnetwork,itcanbejudgedasusefulornotaccordingto the rules. Only information that meets the algorithm’s 4.1.Datasets. In this paper, the daily sales volume data of a certiﬁcation is left behind, while information that does not brand company’s calligraphy exercise products in each match is forgotten through the forgetting gate. It is just a market of a region within three consecutive years were one-in-two-out working principle, but it can solve the long- selected, and the brand company had 78 retail stores in the standing big problems in neural networks with iterative region, and 78 sets of 1096 sales volume time series were operations. It has been shown that LSTM is an eﬀective obtained after statistical collation; one set of sales data and technique to solve the long-order dependence problem and data format are listed in Table 1. *e data in Table 1 can be the generalization of this technique is very high, leading to a used as a random time series for diﬀerent outlets according great variety of possibilities brought about. to the synchronous series sorted from January 1, 2015, to LSTM networks are trained using a backpropagation December 31, 2017. algorithm with the following steps: (1) forward computation By observing the data characteristics of the sales volume ofeachneuronoutputvalue,forLSTM,i.e.,ﬁvevectors i f , of the enterprise’s products and then the above data as a unit t t o , c , and h , which have been described in the previous of weekly statistics sales data, extracting the more repre- t t t section;(2)backwardcomputationoftheerrortermforeach sentative three groups of data for comparative analysis neuron; as with recursive neural networks, the backward shows the sales volume of the enterprise’s products in the propagation of LSTM error terms consists of two directions: period 2015–2017; there are obvious ﬂuctuations; with time, the ﬁrst is the backward propagation along time, i.e., cal- sales data show a more obvious downward trend and a culating the error term at each moment starting from the certain seasonal pattern and the existence of randomness, current moment; the second is propagating the error term mainlyaspartoftheseasonaltrendandtheoveralldeclinein 8 Journal of Mathematics Table 1: Daily sales volume statistics of calligraphy practice forecast interval is speciﬁed as September 2017–December products of a company from 2015 to 2017. 2017,andtheforecastfrequencyis“days”.Inordertopredict the value of sales change in the future period, the original 1 2 3 4 5 6 7 8 9 10 11 12 13 14 ... sales volume time series data was divided into a training set 2015 6 8 11 7 4 5 6 6 5 4 11 6 10 4 ... andatestsetaccordingto9:1asawhole,thetrainingsetwas 2016 15 13 10 9 8 4 5 9 6 13 4 6 4 7 ... inputintothemodelfortraining,andtheProphetmodelwas 2017 13 15 5 5 7 3 8 16 7 5 11 9 5 6 ... used to analyze and predict the results of the sales volume data, which will theoretically decline gradually in ﬂuctua- sales volume of mixed trends, andthe local existence of large tions over time. In order to optimize the update rate of the ﬂuctuation characteristics. LSTM model parameters, the learning rate (Learning Rate) needs to be further controlled. *e Adam algorithm com- bines the advantages of Ada Grad and RMS Prop algorithms 4.2. Model Combinations. In order to fully utilize the ad- to dynamically adjust the learning rate of each parameter by vantages of the Prophet model and LSTM neural network using the ﬁrst-order moment estimation and second-order model,thispaperproposes anoptimal combinedforecasting moment estimation of the gradient. It is an eﬀective gra- model based on Prophet and LSTM neural network for sales dient-based stochastic optimization method, which makes time series forecasting of calligraphy exercise products. In the update of parameters smoother and takes less storage February 2019, Drotar´ et al. open-sourced a set of time series resources. forecasting tools Prophet. Unlike traditional time series forecasting models (ARIMA, etc.), the Prophet model is essentially a curve ﬁt to time series data, while having ex- 4.3. Experimental Results. In order to ﬁnd the optimal cellent adaptability to holiday eﬀects and trend change weight coeﬃcients w and w in the integrated model 1 2 points (change points) in the data, especially robust to Prophet-LSTM model, we take the coeﬃcients w to be 11 missing values, shifts in trends, and a large number of values in [0.0, 1.0] increments, w + w � 1 and then the 1 2 outliers, and is currently mainly applied to traﬃc forecasting corresponding weight coeﬃcients w to be 11 values in [1.0, on Facebook social networking sites. Prophet itself is a 0.0] decrements and multiply the 11 sets of weight coeﬃ- model based on a self-additive model to predict time series cients w and w with the sales volume forecasts of the 1 2 data, the model as a whole consists of growth (trend term), respective models at each moment. After the above process, seasonality (period term), and holidays (holiday term) 3 the11setsofweightedintegratedsalesforecastsareobtained parts superimposed, and the basic form is as follows: as P(t) � g(t) + s(t) + h(t) + ε, (5) Y t􏼁 , i � 1,2, . . . ,11, ti � 1,2, . . . , n. (7) Here, the trend term g(t) is the core component of the In order to ﬁnd the speciﬁc values of the weight coef- whole Prophet model, containing parameters with diﬀerent ﬁcients w and w , the root mean square error equation 1 2 degrees of assumptions and adjusting smoothness, which is (RMSE) and the mean absolute error (MAE) are used as the used to ﬁt the nonperiodic changes in the time series and criteriatoevaluatetheeﬀectoftheProphet-LSTMcombined select the change point (change point) from the data to sales forecasting model for each group of weight coeﬃcients, detectthetrenddirection.Firstly,theProphetmodelandthe andtheexpressions areas follows:wherexistheactualvalue LSTM network model with high prediction accuracy are of product sales in week i. According to Table 2, the root constructedseparatelyforthesalesdata,thepredictionvalue mean square error of the prediction results is minimized of the Prophet model at time t is set as P(t), the prediction when the weight coeﬃcients w �0.6 and w �0.4, and the 1 2 value of the LSTM network model is L(t), t �1, 2, ..., N, and mean absolute error of the prediction results is minimized the two models are assigned dynamic weights w and w , 1 2 when the weight coeﬃcients w � 0.4 and w � 0.6. *e 1 2 respectively. t, this point, deﬁnes the integrated Prophet- resultsshowthatthepredictionresultsoftheProphet-LSTM LSTM combined prediction model as model are improved compared with the Prophet and LSTM models, and its prediction accuracy is generally better than Y(t) � w P(t) + w L(t), w + w � 1, t � 1,2, . . . , N, 1 2 1 2 the other two single prediction models, indicating that the (6) prediction results of the Prophet-LSTM combined predic- wheretisthetimewhenthepredictedvalueappearsandY(t) tion model are relatively eﬀective and more applicable than is the result of summing the predicted data of the Prophet the single prediction model in the sales forecast of this model and LSTM network by weights. *e Prophet-LSTM enterprise. *e weight coeﬃcients corresponding to the neural network sales volume forecasting model is obtained evaluation metrics of the model are shown in Table 3. by calculating the optimal weights composed of the two In order to further verify the application performance of models after integration. *e Prophet model can transform the Prophet-LSTM model, Holt–Winters (exponential the time series by certain functions into combined patterns smoothing model), ARIMA model, Prophet model, LSTM, of diﬀerent time dimensions such as daily, weekly, quarterly, and Prophet-LSTM combined model were used as com- and yearly with overall trends. *e Prophet-LSTM model is parison models in this paper to model the prediction of ﬁrst initialized and empirically set change point �0.15 to another new set of sales data, and the evaluation results of make the growth trend more sensitive to changes. *e each model are listed in Table 3. *e evaluation results show Journal of Mathematics 9 Table 2: Comparison of the performance of diﬀerent models. Evaluation metrics Models RMSE MAE Holt–Winters 35.273 16.976 ARIMA 23.457 13.650 Prophet 3.465 2.143 LSTM 5.746 3.254 Prophet-LSTM (w �0.6, w �0.4) 2.645 1.645 1 2 Prophet-LSTM (w �0.4, w �0.6) 7.103 3.635 1 2 Table 3: *e weight coeﬃcients correspond to the evaluation metrics of the model. w 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.0 w 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 RMSE 3.700 3.427 3.274 3.153 3.059 3.022 3.010 3.050 3.139 3.253 3.490 MAE 2.514 2.380 2.300 2.274 2.269 2.297 2.351 2.453 2.524 2.605 2.778 that Prophet and LSTM single models have comparable forecasting model combining two single forecasting models forecasting performance, and both are signiﬁcantly better is designed, on which new forecasting models can be in- than Holt–Winters and ARIMA classical time series models; troduced for further combination in the future to pool the Prophet-LSTM combined forecasting model has further advantages of more models and thus obtain more accurate optimized the forecasting accuracy based on the single forecasting results. Besides, further experiments can be done model and has the best forecasting eﬀect. on the parameter preferences of LSTM neural networks in the follow-up work to seek a better modeling method. In the future, in order to analyze the sales volume time series data 5. Conclusion more deeply, further research on the inﬂuencing factors of sales volume can be targeted. Calligraphy is the art of writing Chinese characters. *e Chinese people love calligraphy and practice it. In a general sense, the art of calligraphy is the art of writing; writing can Data Availability onlymeet theaestheticrequirementsoftheartofcalligraphy *e data used to support the ﬁndings of this study are if it is pursued in an aesthetically pleasing way and subli- available from the corresponding author upon request. mated to art. From the practice of calligraphy education, learning calligraphy has many functions. *e market for calligraphy practice products is vast, but in the face of a Conflicts of Interest complex and dynamic market, it is important to predict the *e author declares that he has no conﬂicts of interest. market demand for calligraphy practice products in a timely and accurate manner, to analyze the characteristics and patternsofsalesvolumedata,andtoimprovetheaccuracyof References salesvolumeforecasting,sothatcalligraphypracticeproduct [1] S. Lin and X. Ban, “Study on cultural and creative design of enterprises can develop timely and eﬀective marketing traditional chinese calligraphy tools in digital era,” in Pro- strategies.*ispaperproposesacombinedmodelprediction ceedings of the 6th International Conference on Arts, Design method based on recursive neural network model based on and Contemporary Education (ICADCE 2020), pp. 427–430, the pattern of sales volume time series data of a calligraphy Jaipur, India, Febraury 2021. practice product enterprise, constructs a weighted Prophet- [2] A. Ahmad Zarnuji, H. Hanif Amrulloh, and I. N. Isnaini Nur LSTM combined prediction model, and conducts compar- Azizah, “Utilization of rice husk waste for paper raw materials ison experiments with the model under each weight coef- as an Arabic calligraphy media,” Engage: Jurnal Pengabdian ﬁcient, the single model before combination, and two Kepada Masyarakat, vol. 3, no. 1, pp. 43–54, 2019. classical time series models. *e experimental results show [3] Y. Dai, “Research on the application of computer multimedia in calligraphy education,” Journal of Physics: Conference Se- that the forecasting performance of the Prophet and LSTM ries, vol. 1915, no. 3, Article ID 032019, 2021. neuralnetworkmodelsissigniﬁcantlyhigherthanthatofthe [4] L. Xiang and L. L. S. Gee, “*e importance of lines: the typical time series forecasting models, and the combined modern calligraphy of wang dongling and qiu zhenzhong,” in Prophet-LSTM model has better forecasting performance. Proceedings of the 2nd International Conference on Interdis- Overall,basedontheveriﬁcationthattheProphetmodeland ciplinary Arts & Humanities (ICONARTIES) 2020, Yogya- the LSTM model have obvious advantages in forecasting the karta,IndonesiaAvailable atSSRN3800559,2021,Yogyakarta, sales volume of calligraphy practice products, this paper Indonesia, March 2021. further improves the model prediction performance by [5] R. Long, C. Sun, and H. Pan, “Research on the design and weighted combination, maximizing the advantages of both application of persuasive game in calligraphy learning,” in prediction models. In this paper, only a combined ProceedingsoftheInternationalConferenceonAppliedHuman 10 Journal of Mathematics Factors and Ergonomics, pp. 473–479, San Diego, CA, USA, July 2020. [6] Y.Li,“CalligraphyEducationinTeachingChineseasaSecond Language,” ;e Palgrave Handbook of Chinese Language Studies, vol. 5, pp. 1–29, 2020. [7] P. Lyu, X. Bai, C. Yao, H. Tengteng, and L. Wenyu, “Auto- encoder guided gan for Chinese calligraphy synthesis,” in Proceedings of the 2017 14th iapr international conference on document analysis and recognition (icdar), pp. 1095–1100, Kyoto, Japan, November 2017. [8] M. R. Rosenberger, C. K. Dass, H.-J. Chuang et al., “Quantum calligraphy: writing single-photon emitters in a two-dimen- sional materials platform,” ACS Nano, vol. 13, no. 1, pp. 904–912, 2019. [9] H. S. R. Kao, M. Xu, and T. T. Kao, “Calligraphy, psychology and the confucian literati personality,” Psychology & Devel- oping Societies, vol. 33, no. 1, pp. 54–72, 2021. [10] Y.-W. Zhang, Y. Chen, H. Liu, Z. Ji, and C. Zhang, “Modeling Chinese calligraphy reliefs from one image,” Computers & Graphics, vol. 70, pp. 300–306, 2018. [11] Y. Weng, X. Wang, J. Hua, H. Wang, M. Kang, and F.-Y. Wang, “Forecasting horticultural products price using ARIMAmodeland neuralnetworkbased on a large-scaledata set collected by web crawler,” IEEE Transactions on Com- putational Social Systems, vol. 6, no. 3, pp. 547–553, 2019. [12] A. Di Piazza, M. C. Di Piazza, G. La Tona, and M. Luna, “An artiﬁcial neural network-based forecasting model of energy- related time series for electrical grid management,” Mathe- matics and Computers in Simulation, vol. 184, pp. 294–305, [13] K. Muralitharan, R. Sakthivel, and R. Vishnuvarthan, “Neural network based optimization approach for energy demand prediction in smart grid,” Neurocomputing, vol. 273, pp. 199–208, 2018. [14] H. Eskandari, M. Imani, and M. P. Moghaddam, “Convolu- tional and recurrent neural network based model for short- term load forecasting,” Electric Power Systems Research, vol. 195, Article ID 107173, 2021. [15] I. Matino, S. Dettori, V. Colla, V. Weber, and S. Salame, “Forecasting blast furnace gas production and demand through echo state neural network-based models: pave the way to oﬀ-gas optimized management,” Applied Energy, vol. 253, Article ID 113578, 2019. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Mathematics Hindawi Publishing Corporation http://www.deepdyve.com/lp/hindawi-publishing-corporation/recursive-neural-network-based-market-demand-forecasting-algorithm-for-9JqSr4Oppp

Loading next page...

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

Publisher: Hindawi Publishing Corporation
Copyright: Copyright © 2022 Yi Xue. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ISSN: 2314-4629
eISSN: 2314-4785
DOI: 10.1155/2022/8086789
Publisher site: See Article on Publisher Site

Abstract

Hindawi Journal of Mathematics Volume 2022, Article ID 8086789, 10 pages https://doi.org/10.1155/2022/8086789 Research Article Recursive Neural Network-Based Market Demand Forecasting Algorithm for Calligraphy Practice Products Yi Xue Jilin University of Architecture and Technology, Jilin, Changchun 130000, China Correspondence should be addressed to Yi Xue; 2016121641@jou.edu.cn Received 17 November 2021; Revised 30 November 2021; Accepted 6 December 2021; Published 5 January 2022 Academic Editor: Naeem Jan Copyright © 2022 Yi Xue. *is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Intoday’ssociety,calligraphy,whichreﬂectsone’sbasicwritingskills,isbecomingmoreandmoreimportanttopeople.Peopleare inﬂuenced by calligraphy in their studies, work, etc. Improving calligraphy writing skills has become one of the key directions for developing one’s abilities at this stage. As an important means of improving writing skills, calligraphy practice products are attractingmoreandmoreattentionandpurchases.Inparticular,inrecentyears,asthemarketeconomyhasdevelopedinadeeper direction, people’s demand for calligraphy practice products has diversiﬁed and calligraphy practice product companies have launched a varietyof productsto meetthe public’scalligraphypracticeneeds in ordertoadapt tothe reality of consumer demand. However, with the development of the Internet culture industry and inﬂuenced by objective factors such as school holidays and seasons, the current market demand for calligraphy practice products is rapidly and dynamically changing, making market changes diﬃcult to grasp and leading to poor sales, which directly aﬀects the proﬁts of calligraphy practice product-related companies. *e artiﬁcial intelligence neural network method realizes the nonlinear relationship between the input and output of sampledatathroughtheself-learningabilityofeachneuronandhasacertainnonlinearmappingabilityinprediction,whichplays a great role in the market demand prediction of many commercial products. Based on this, this paper proposes a recursive neural network-based algorithm topredict the future demandand development trend of calligraphy practice products through extensive and in-depth research, so as to provide positive and beneﬁcial guidance for enterprises’ future production and sales. vast sea of Chinese culture. *e art of calligraphy has always 1. Introduction been respected by the nation as an elegant, erudite, and China has been a cultural power since ancient times, and nurturing art form. *e art of Chinese calligraphy is a after more than a thousand years of historical evolution, the manifestation of Chinese culture and is one with the spirit of major ancient Chinese civilizations have long inﬂuenced the Chinese nation. *e spirit of Chinese culture is the unity each other and merged. Today, a China with a splendid of heaven and man, the valuing of harmony, and the culture stands in the East of the world with its colorful shunningofcenter.*eChinese culture’svalueof“harmony is precious” is perfectly embodied through the beauty of cultural elements. An important part of Chinese culture is writing, and the diﬀerent aesthetic forms of writing form the harmony in the art of calligraphy [2]. Calligraphy practice art of calligraphy [1]. *e art of Chinese calligraphy, with its products collection is shown in Figure 1. long and unbroken history, has attracted generations of Chinese calligraphy, with its ability to express the calligraphers to cultivate and tread the paths of the ancients, richness and complexity of human thought and emotion in to experience the essence of the culture that has been the simple shape of lines, has great aesthetic value. *e study precipitated for thousands of years, and on the other hand, of Chinese calligraphy is a great help in understanding to read ten thousand books, travel ten thousand miles, Chinese art and culture. Calligraphy exists alongside the capture the world, cultivate their bodies, and express it in written word and is concerned with brushwork, strokes, and theirbrushstrokes,creatingastyleofcalligraphyappropriate strokes of meaning. With the help of Chinese characters, to the times. Together with other forms of art, they form a calligraphy is an expression of sentiment and sentimentality 2 Journal of Mathematics Figure 1: Calligraphy practice products collection. and has a strong national character and high artistic taste. cultivate the body and nourish the mind. Calligraphy is indeed a great source of health beneﬁts, allowing the whole Chinese calligraphy has been steeped in thousands of years and is ubiquitous in Chinese life. It is closely related to bodytomove,thewaistandarmstrengthandeyestrengthto be harmonized, and the rigidity and ﬂexibility to be com- residential culture, landscape culture, educational culture, religious culture, political culture, coin culture, folk ritual bined with movement and stillness, as well as strengthening customs, and the arts of painting, architecture, dance, arts, the body and brain, eliminating fatigue, regulating the and crafts [3]. As shown in Figure 1, Chinese calligraphy is a psyche,andquietingthemind.Calligraphersandcalligraphy unique visual art that still exudes a fascinating artistic charm education work have also become a highly respected pro- after thousands of years. *e artistic beauty of Chinese fession in society. *e Chinese Calligraphers Association, calligraphy is composed of two elements: form and spirit. which consists of national calligraphers, seal carvers, cal- *e beautiful and distinct artistic eﬀects and infectious ligraphy theorists, calligraphy educators, and calligraphy power of calligraphy reﬂect a unique national style. *e activity organizers and management workers, has become a national professional organization and a group member of composition of traditional painting and calligraphy requires the author not only to make a subjective and rational the Chinese Federation of Literary and Art Circles. Callig- judgment but also to take the various elements of the picture raphy enthusiasts have relatively good social recognition. and to achieve a dialectical unity between the whole and the *is intangible social recognition and support provide great parts so that the picture can achieve balance and harmony, assistance in the spread, learning, and development of which can be regarded as the earliest theory of composition calligraphy [5]. in Chinese painting [4]. *e aesthetic value of the art of Chinese calligraphy has always been an excellent tra- calligraphy has many applications in home decoration, ditionalculturethatweareproudof,withthousandsofyears where its content has a subtle eﬀect on the human spirit, of history. As life has improved, many parentshave begun to gaining knowledge and culture, drawing inspiration and pay attention to the cultural development of their children. strength. At the same time, calligraphy with its unique Learning calligraphy can lay the foundation for children’s writing and can improve their overall quality. China’s means of art life and service life is widely used in the design of patterns for product packaging. *e art of Chinese cal- contemporary design started late, and so did the stationery ligraphy,whichonthesurfaceappearstobeverycommon,is industry’sdesignsystem,withalargeproportionofdomestic nothing more than taking a brush and dipping it in ink to stationery manufacturers borrowing or directly copying write Chinese characters on paper (of course, other writing foreign stationery products in the early days. However, as a carriers are sometimes used), butit is this extreme simplicity local cultural product, it is diﬃcult to ﬁnd a mature product and richness of variation that has made calligraphy less from abroad that is worth learning from. With the rapid attractive in contemporary times due to the advancement of development of the Internet culture industry, calligraphy modern science and technology, especially the spread of practice products are now combined with modern design to computers and the reduction of handwriting. It has become create new developments that shine in a diﬀerent light [6]. one of the most popular art forms in contemporary China, As a representative of traditional culture, the calligraphy practice product industry is also facing unprecedented with the largest number of participants and the widest audience. Calligraphy is widely recognized for its ability to impact, challenges, and opportunities in the Internet era, Journal of Mathematics 3 and it is worthwhile to deeply explore the artistic value of First, consumables such as ink, paper, and inkstone are calligraphy and promote it, thus creating economic value relatively inexpensive compared to other hobbies, such as and prosperity for the calligraphy industry. In recent years, musical instruments and photography equipment. *e vast with the rapid development of computer technology, the majority of consumers can aﬀord them. Beginners can Internet is deeply integrated with various traditional in- basically get all the materials they need to learn calligraphy dustries. *e Internet economy,represented by the booming for under $1,000. With the rise of productivity and the e-commerce, has become an important engine of economic economy, the cost of consumables for calligraphy practice is development. *is will provide great development potential no longer an issue that hinders the spread of calligraphy and broader development space for the rise of the Internet learning. Secondly, the calligraphy training industry is also economy. *e Internet has changed people’s lifestyles and in full swing with economic development. As more and brought more possibilities for consumption upgrading and more families reach the well-oﬀ level and have a relatively industrial development. *e further integration of the In- goodincome,parentshavethefundsandabilitytosendtheir ternet with traditional industries has changed the way the children to calligraphy training courses to learn calligraphy. *e cost of calligraphy education is in the middle of the original industries produce,sell, andcommunicate, bringing new opportunities for development. As a typical represen- range of hobby learning costs and is becoming more and tative of traditional industries, the calligraphy practice more acceptable to parents and families. *e good economic product industry is also integrating Internet technology, development and the increase in consumption power have quietly undergoing the biggest change in thousands of years. created a good development environment and opportunity Artiﬁcial intelligence is a new technical science that studies for the calligraphy training industry. In addition, economic and develops theories, methods, technologies, and appli- development has led to the development of the Internet and cation systems used to simulate, extend, and expand human online education, and payment for knowledge has become intelligence. Artiﬁcial intelligence-based market demand theconsensusofthetimes.Onlineeducationandonlinepaid forecasting method refers to the method of using artiﬁcial courses in calligraphy are also being accepted by more and intelligence techniques, mainly machine learning and deep more adult learners, all of which have brought great pros- perity and opportunities to the calligraphy practice products learning, etc., to simulate industry experts for market de- mand forecasting, which diﬀers from fundamental analysis market. Practitioners of calligraphy practice products have and technical analysis methods in terms of modeling also enjoyed the dividends of economic development, and methods, and focuses on the use of artiﬁcial intelligence calligraphy, an important art form, has gradually become a techniques to uncover potential connections between data leisurely and elegant way of life for the general public after from various sources in the market and market movements. the majority of people have solved the problem of food and In order to adapt to the new environment of the Internet clothing [7]. culture industry, it is necessary to design and study new Withthedevelopmentofsocietyandtheadvancementof intelligent ﬁeld demand forecasting algorithms for the cal- technology and culture, the calligraphy industry has also ligraphy practice product market using artiﬁcial intelligence seen unprecedented growth. Various calligraphy practice techniques. products have become abundant, the four treasures of In this paper, we start by presenting related works in calligraphy supplies continue to innovate, and new teaching Section 2; then, in Section 3, we discuss an artiﬁcial intel- models and online classes have become popular with the ligence neural network method that has certain nonlinear advancementofvideoandlivestreamingtechnology.Online mapping ability in prediction; and for the concerned shopping, online payments, and interpersonal trust have method,wedoexperiments inSection4andevaluationsand taken calligraphy practice products from oﬄine to online. It ﬁnally put a conclusion in Section 5. can be said that the calligraphy industry has also enjoyed the great convenience of the Internet and technological ad- vances, and the dissemination of calligraphy has become 2. Related Work easier and more convenient. However, the calligraphy 2.1. Calligraphy Practice Products Market Analysis. As an market still has its limitations [8]. Calligraphy is a niche important part of art and culture, calligraphy has gradually industry with a small group of learners, and although cal- ligraphy can be found everywhere in life, it is still a minority moved from the study to the marketplace, forming the basic chain of industrialization, the emergence of specialized of people who actually learn and come into contact with calligraphy, so the small consumer base and low degree of calligraphic art dealers and collectors, and the involvement of merchants in the operation of calligraphic artwork that consumer demand are the biggest obstacles to building a calligraphy industry chain. Compared to the entertainment has to some extent further guided the development of calligraphy in the direction of industrialization. At the same and leisure industries, such as movies, music, dance, and time, the development of a related industrial chain plus the food, which have a high degree of popularity, calligraphy is framing of calligraphic works and the trading of the Four still relatively inactive. Calligraphy is a quiet art, limited to Treasures of calligraphy market gradually took shape. As the the line and visual appreciation, lacking auditory and gus- tatory stimulation, with limited impact and limited inﬂu- national economy is developing rapidly and people’s con- sumption levels are increasing, the art of calligraphy is also ence, and with the popularity of computer fonts, the practicality of calligraphy is even weaker. Many people are seeing better opportunities for development. People have more time and money for calligraphy-related consumption. not involved, making market activity and demand 4 Journal of Mathematics servants, etc.). K12 calligraphy post is the largest segment of lukewarm. *e cycle of production (creation), sales, and consumption of calligraphy are slow, circulation is tiny, age the entire calligraphy post industry, accounting for more than 60% of the market. At present, China’s character post groups are large, and innovation is slow. Calligraphy products are also nondepleting; many calligraphy products industry is mainly led by several well-known calligraphers, can be used for years, and many calligraphy paintings can be and there are mainly two types of calligraphy practice preserved for years, which makes the replacement of new character post usage forms, one is the tracing character post, calligraphy products slow, and low consumption and low and the other is the water writing character post. *eir attrition make it diﬃcult for demand for calligraphy common drawback is the lack of interaction, the use of a productstorise.Calligraphyisanelegant,high-end,spiritual single mode, the use of boring, and not being able to write a variety of fonts. It is based on thelack of interaction between art with a high skill threshold, requiring a certain level of practice, cultural heritage, and knowledge to resonate, love, the existing handwriting posters and the practitioner where the laser projection technology and VBK keyboard tech- and appreciate. Calligraphy also has regional and cultural limitations, limited to a few countries such as China and nology are integrated into the interactive laser projection calligraphy practice product, which has the advantages of Japan, and from an international perspective, it lacks global demand and circulation. All of this has hindered the rapid easy writing, strong interactivity, and assistance in cor- development of the calligraphy market. *e calligraphy recting posture. It has three practice modes. First, learning industry is a more stable, long-term lukewarm state of mode: the laser emitter projects the trajectory and strokes of development, with an overall stable Baidu search index and calligraphy on the paper, visually demonstrating the writing information index with a slight increase. Calligraphy is a technique; second, copying mode: the laser emitter projects traditional culture of thousands of years, and even with the the outer outline of the font on the paper, and the user can directly copy the practice; third, practice mode: the laser impact of the Internet and artiﬁcial intelligence, it will continuetobe relativelystableandwillnotexperiencemajor emitterprojectsthemetergridonthepaper,andtheusercan practice writing, and after the writing is completed, the ups and downs [9]. *e calligraphy industry is inactive, and the calligraphy auction industry is relatively sluggish, with product’s intelligent scoring function will rate and voice correct the user’s writing. After writing, the product’s in- fewer inﬂuential calligraphy works and calligraphy masters, although precious calligraphy works can fetch high prices. telligent scoring function will rate the user’s writing and *ecalligraphyindustryisindisarray,withsmallcalligraphy voicecorrection.Atthesametime,thesystemissetupwitha practice product manufacturers stealing and evading taxes, library of diﬀerent types of fonts, so users can choose dif- andcalligrapherswithlowtaxcapacityorconductingprivate ferent types of fonts for practice. *e interactive laser transactions without paying taxes. Numerous small and projection calligraphy practice product makes up for the scattered calligraphy training institutions are popping up, shortcomings of the existing calligraphy posters while enriching the functions of the calligraphy practice tool to and paying taxes is not regulated enough. *e state does not havemuchrevenuetospendonthecalligraphyindustry,and better meet the needs of consumers. In terms of product functionality, the interactive laser projection calligraphy there is no major investment. In the calligraphy industry analysis, the competition in the same industry for sales of practice product has a great advantage in that it allows for calligraphy products also needs to be mastered. Oﬄine good human-computer interaction and can meet the re- competition can be judged by market research; whether quirements of users with diﬀerent fonts, while being able to there are local calligraphy training institutions, calligraphy detectandcorrectthewordswrittenbytheuser,preservinga galleries, and calligraphy stores, etc., and how the customer realistic writing experience. In terms of functionality, this ﬂow and sales are can be investigated. Competition on product is the equivalent of a calligraphy postcard with online e-commerce platforms can be obtained through big thousands of fonts in one and a calligraphy teacher at the data analysis tools on e-commerce platforms, allowing for a same time, providing a one-stop service for learning, practicing, and correcting calligraphy, which is not possible clear, accurate, real-time grasp of competition in the same industry. *e Internet and big data provide tremendous and with the calligraphy practice products currently available on the market. *e interactive projection calligraphy practice convenient help for marketing analysis and business deci- sions. In conclusion, calligraphy practice products have product has advanced technical support to ensure that its enjoyed the dividends of the Internet technology era, but function is achieved. *e product uses the currently mature there are also great limitations and space, which requires the laser projection technology and VBK keyboard technology, joint eﬀorts of every calligraphy lover and calligraphy in- with a built-in laser emitter to project the interface infor- dustry practitioners to pioneer and innovate and constantly mation on the surface of any paper and then track the ﬁnger explore to make it prosperous and strong. movements through infrared technology to ﬁnally complete theacquisition of inputinformation. High-tech productsare Calligraphy practice is often easy to start and diﬃcult to become a master, or it is easy for beginners to practice a trend for future product development, and consumer enthusiasm for high-tech products is always growing. When calligraphy through text posters. China’s calligraphy post industry according to user classiﬁcation is mainly divided high-tech elements are incorporated into a product, then the product can stand out among similar products. *e inter- into the K12 calligraphy post market (the main users are kindergarten to high school), adult calligraphy post market, active laser projection calligraphy practice product is pre- and professional calligraphy post market (the main users ciselytheintegrationoflaserprojectiontechnologyandVBK with certain career characteristics, such as doctors, civil keyboard technology that makes the product stand out Journal of Mathematics 5 that characterize trends. In daily applications, RNN models among calligraphy practice products. With the increasing popularity of high-techproducts, interactive laser projection are widely used in tasks such as speech recognition and machine translation, bringing convenience to people’s lives. calligraphy practice products will be favored by many consumers. According to the “Guideline for Primary and In the biomedical ﬁeld, sequence data are even more diverse Secondary School Calligraphy Education” published on the and large in scale, in which RNNs are widely used for tasks website of the Ministry of Education, elementary school such as health monitoring, auxiliary diagnosis, and infor- grades 3 to 6 are required to schedule one class period per mation extraction, making great contributions to medical weekforbrushwritingstudy.Accordingtostatisticsreleased development as well as human health. Most biomedical by the National Bureau of Statistics on February 24, 2015, it signals are time-varying (time-variant) nonlinear signals, can be seen that 93.6 million students were enrolled in including electrocardiogram (ECG), electroencephalogram generalelementaryschoolsnationwidein2015.*ereissuch (EEG), and electromyogram (EMG). Based on ECG signals, RNN can be used for arrhythmia diagnosis. Based on EEG a huge group of calligraphy practitioners among children. *e latest survey results released by Regus, a world-re- signals, RNN can be used for sleep signal classiﬁcation. In addition, in the brain-computer interface, RNN can be used nownedoﬃcesolutionsprovider,saidthatthestresssuﬀered by oﬃce workers in mainland China in the past year ranked to decode EEG signals and improve the grasping accuracy of ﬁrst in the world. Nowadays, oﬃce workers’ life is neural prostheses. In recent years, the multimodal (multi- accelerated and their work life is irritating; they are also modal) biomedical signal utilization approach has become a eager to seek a new way to reduce stress. Practicing cal- trend. For example, RNN can simultaneously utilize EEG ligraphy can cultivate the mind, practice qi, and beneﬁt the signals as well as functional near-infrared spectroscopy (f brain,nurturerespect,andgreatlyreducethestressfromlife. NIRS) for epilepsy monitoring. Text in biomedicine is even more complex and diverse, and it is natural that RNNs can With the development of society and the aging of the population, more and more elderly people are experiencing perform natural language processing (NLP) on such data [12]. For example, RNN-based biomedical named entity “retirement syndrome” because they do not know how to organize their lives after retirement. In addition to con- recognition (BNER) technology can help researchers extract usefulinformationfrommassivebiomedicaltexts.Inclinical sideringretirementasarest,theelderlyshouldalsomaintain a certain level of tension so that life does not become boring. practice, medical images including ultrasound images, CT *is shows that there is a huge demand for calligraphy images, and MRI images are an important basis for disease practice tools. In this regard, we did an interview survey, by diagnosis. Based on ultrasound image sequences, RNN interviewing some parents, teachers, students, oﬃce models can signiﬁcantly improve the detection accuracy of workers, and retired elderly people, the survey results show prostate cancer. In computer-aided detection (CAD), RNN thatprimaryand secondaryschoolstudentshavetheneed to can improve the accuracy of early breast cancer detection using current X-ray scans as well as previous scans. At the practice calligraphy, and it is very important for the de- velopment of students. Working people have to sign con- microscopic scale, RNNs have also been successful in the task of target tracking in time-lapse cell image sequences. In tracts, agreements, examinations for titles, etc., and they need to improve their writing skills urgently, and the de- the clinical setting, electronic health record (EHR) data mand for word practice is also great. Secondly, retired se- containssequencesofdiﬀerentmodalities,whichistypicalof niors: surveys show that many retired seniors prefer to read multimodal data. Based on EHR, RNN can perform disease and write then to play cards and watch TV and other similar prediction and also predict the risk of readmission of dis- activities. So according to the survey results, the target charged patients over a period of time, providing the pos- market for calligraphy practice products should be basically sibility of early initiation of targeted intervention programs located in primary and secondary school students, oﬃce for patients at risk of readmission, eﬀectively reducing the workers, and retired seniors [10]. probability of readmission, and controlling the cost of care. In recent years, dynamic treatment recommendation sys- tems based on large-scale electronic health records have emerged as the key to the successful improvement of actual 2.2. Recursive Neural Networks. Artiﬁcial neural network (ANN) has been named deep learning at the time of rapid clinicaloutcomes.However, thesesuccessesdonotconceala development of computer storage, communication, and series of problems that arise during the backpropagation of computing power and has set oﬀ the third artiﬁcial intel- error signals in the depth structure when training RNN ligence boom in the world’s history with unprecedented models with gradient descent, mainly including the van- scale. “Deep” gives neural networks the ability to abstract ishing and exploding gradient problem and the gradient information in a hierarchical way, giving them the “intui- conﬂict. *e former is the most common problem in the tion” that people have [11]. Overnight, many ﬁelds that had general feed-forward neural system. *e former is also beenﬁrmlydominatedbyﬁne-grainedmathematicalmodels frequently mentioned in general feed-forward neural net- work (FNN) research [13]. Before the popularization of let go of their previous obsession with equations and their prejudice against the inscrutable black box and jumped into Rectiﬁed Linear Unit (ReLU), FNNs with deeper structures were often diﬃcult to train due to these problems (especially the deep learning bandwagon. In this transformation, a recursive neural network (RNN) is responsible for modeling the gradient vanishing problem). Unlike FNNs, the depth sequential data. Common serial data include signals, text, structure of RNNs is inherent, because even an RNN model video image sequences, and serial data such as stock prices that contains only a single layer of hidden units in space 6 Journal of Mathematics becomesverydeepwhenitisexpandedintime.RNNmodels Forget Gate have the ability to remember, are turing complete, and can Self-recurrent theoretically learn dependencies at any time interval in a connection sequence. However, the long-time memory capability of RNNs is not reliable due to the aforementioned problems. Memory cell Memory cell Speciﬁcally, it is very diﬃcult to rely on gradient descent to Input output cell train RNN models to learn long-time dependencies, and early RNN models with simple structures are often deep in the quagmire of gradient disappearance. *e Long Short- Input Gate Term Memory (LSTM) proposed at the end of the last Output Gate century largely alleviated the gradient disappearance Figure 2: Internal structures of neurons in recurrent neural problem, which made RNNs shine in this century’s boom, networks. and the LSTM structure itself became the standard con- ﬁguration of most RNN models. However, LSTM has not fundamentally changed the dilemma that the RNN family is created to address the need for sequential inputs. Long in, and gradients still vanish in the face of longer time series. Short-Term Memory (LSTM) is essentially a speciﬁc form of On the other hand, the gradient explosion problem has recursive neural network (RNN). Inaddition to containing a neverbeenproperlysolved.Atthetimeofwritingthispaper, memory module, LSTM solves the RNN long-term de- the third wave of neural networks is still going on. *e pendency problem by adding a threshold (Gated RNN) to amount of data available for learning by various neural solve the problem that the LSTM adds input gate, output models is increasing dramatically, and new sequence data gate, andforget gate tothe basic structure of RNN [15]. Each and their corresponding learning tasks are emerging. *e of these three logic control units is connected to a multi- long-time dependencies embedded in the sequence data are plication element (see Figure 2), and the input and output of becoming richer, longer in time span, and more obscure in the information ﬂow and the state of the memory cell are presentation, and capturing these long-time dependencies controlled by setting the weights at the edges where the eﬃciently and accurately is the key to dealing with such memory cell of the neural network is connected to the other problems. On the other hand, neural network models are parts. *e speciﬁc structure is shown in Figure 2. increasing day by day. Increasing the network size by To note, the input gate denoted as i controls whether stacking model layers and increasing link lengths usually information ﬂows into the memory cell whereas the forget acceleratesthedecline ofmodelmemory.Inthiscontext and gate controls whether the information in the memory cell at historical trend, it is urgent to investigate the RNN archi- the previous moment accumulates into the memory cell at tecture and related algorithms with long-time memory ca- the current moment, and it is denoted asf . *e output gate pability. *e method proposed in this paper is the market denoted as o controls whether the information in the demand prediction of calligraphy practice products with memory cell at the current moment ﬂows into the currently long- and short-term memory network architecture. hidden state h . We recall also the cell “memory cell”, which represents the memory of the neuron state and makes the 3. Method LSTM cellhavetheabilitytosave,read,reset, andupdatethe long history information, denoted as c . At moment t, the *e Internal structure of neurons in recurrent neural net- LSTM neural network is deﬁned by the following equation: works is shown in Figure 2. f � sigmoid􏼐W · 􏼂h , x 􏼃 + b 􏼑, A neural network is a model with strong learning ability, t f t−1 t f especially eﬀective in dealing with pattern recognition, in- i � sigmoid W · 􏼂h , x 􏼃 + b 􏼁, t i t−1 t i telligent control, and other problems. In recent years, with o � sigmoid W · h , x + b , the booming development of computer science and tech- 􏼂 􏼃 􏼁 t o t−1 t o (1) nology and hardware devices, neural networks have grad- 􏽥 c � tanh W · 􏼂h , x 􏼃 + b 􏼁, t c t−1 t c ually become a hot research topic in the ﬁeld of artiﬁcial c � f ∗ c + i ∗ c , intelligence [14]. However, Full Connected Neural Network t t t−1 t t (FCNN) has the limitation that the nodes of the same layer h � o ∗ tanh c􏼁 . t t t are not connected to each other, and FCNN cannot be used when the information of the previous moments of the se- *e recursive neural network timing diagram is shown quence is needed. Since FCNNs cannot share features be- in Figure 3. tween diﬀerent locations of a sequence, they can only handle In addition to the previously mentioned i f , o , and c , t t t t one input individually, i.e., there is no relationship between W represent the recursive connection weights of their the previous input and the next input, and they cannot corresponding gates, and sigmoid and tanh are the two handle inputs that are related back and forth in time or activation functions. In the training process of the LSTM space. However, many learning tasks need to deal with neural network, ﬁrstly, the data features at moment t are sequential information, such as time series prediction and input to the input layer and the results are output through task-based dialogues, which require the model to learn from the excitation function. *e output results, the output of the sequential inputs. Recursive Neural Network (RNN) was hidden layer at moment t−1, and the information stored in Journal of Mathematics 7 C C C C C t-n t-n+1 t t+1 t+2 c c c c h h h h h t-n t-n+1 t t+1 t+2 h h h h x x x x t-n t-n+1 t t+1 Figure 3: Recursive neural network timing diagram. the cell unit at moment t−1 are input into the nodes of the z � σ V · s , (2) t t LSTM structure, and the data are output to the next hidden layer or output layer through the processing of input gate, s � δ U · x + W · s , (3) t t t−1 output gate, forget gate, and cell units, the results of the nodesoftheLSTMstructureareoutputtotheneuronsofthe where V is the weight matrix of the output layer; σ is the output layer, and the calculation of backpropagation error output layer activation function; U is the weight matrix of updates each weight. *e overall structure is shown in the input x; W is the weight matrix of the hidden layer state Figure 3. s at moment t−1 as the input at moment t; g is the hidden t−1 A simple recursive neural network consists of an input layeractivationfunction.Loopingequation(3)intoequation layer, a hidden layer, and an output layer. For a given input (2), we get sequence x � [X , X , . . . , X ], at moment t, the hidden 1 2 t layer state is s and the output value is z , as follows: t t z � V∗δ U∗ x + W∗δ U∗ x + W∗δ U∗ x + W∗δ U∗ x + · · ·􏼁􏼁􏼁􏼁. (4) t t t−1 t−2 t−3 *e LSTM diﬀers from the RNN in that it adds a one layer up; (3) calculating the gradient of each weight “processor” to the algorithm to determine whether the in- according to the corresponding error term. formation is useful or not. A cell has three gates, called input gate,forgetgate,andoutputgate.Whenamessageentersthe 4. Experimentation and Evaluation LSTMnetwork,itcanbejudgedasusefulornotaccordingto the rules. Only information that meets the algorithm’s 4.1.Datasets. In this paper, the daily sales volume data of a certiﬁcation is left behind, while information that does not brand company’s calligraphy exercise products in each match is forgotten through the forgetting gate. It is just a market of a region within three consecutive years were one-in-two-out working principle, but it can solve the long- selected, and the brand company had 78 retail stores in the standing big problems in neural networks with iterative region, and 78 sets of 1096 sales volume time series were operations. It has been shown that LSTM is an eﬀective obtained after statistical collation; one set of sales data and technique to solve the long-order dependence problem and data format are listed in Table 1. *e data in Table 1 can be the generalization of this technique is very high, leading to a used as a random time series for diﬀerent outlets according great variety of possibilities brought about. to the synchronous series sorted from January 1, 2015, to LSTM networks are trained using a backpropagation December 31, 2017. algorithm with the following steps: (1) forward computation By observing the data characteristics of the sales volume ofeachneuronoutputvalue,forLSTM,i.e.,ﬁvevectors i f , of the enterprise’s products and then the above data as a unit t t o , c , and h , which have been described in the previous of weekly statistics sales data, extracting the more repre- t t t section;(2)backwardcomputationoftheerrortermforeach sentative three groups of data for comparative analysis neuron; as with recursive neural networks, the backward shows the sales volume of the enterprise’s products in the propagation of LSTM error terms consists of two directions: period 2015–2017; there are obvious ﬂuctuations; with time, the ﬁrst is the backward propagation along time, i.e., cal- sales data show a more obvious downward trend and a culating the error term at each moment starting from the certain seasonal pattern and the existence of randomness, current moment; the second is propagating the error term mainlyaspartoftheseasonaltrendandtheoveralldeclinein 8 Journal of Mathematics Table 1: Daily sales volume statistics of calligraphy practice forecast interval is speciﬁed as September 2017–December products of a company from 2015 to 2017. 2017,andtheforecastfrequencyis“days”.Inordertopredict the value of sales change in the future period, the original 1 2 3 4 5 6 7 8 9 10 11 12 13 14 ... sales volume time series data was divided into a training set 2015 6 8 11 7 4 5 6 6 5 4 11 6 10 4 ... andatestsetaccordingto9:1asawhole,thetrainingsetwas 2016 15 13 10 9 8 4 5 9 6 13 4 6 4 7 ... inputintothemodelfortraining,andtheProphetmodelwas 2017 13 15 5 5 7 3 8 16 7 5 11 9 5 6 ... used to analyze and predict the results of the sales volume data, which will theoretically decline gradually in ﬂuctua- sales volume of mixed trends, andthe local existence of large tions over time. In order to optimize the update rate of the ﬂuctuation characteristics. LSTM model parameters, the learning rate (Learning Rate) needs to be further controlled. *e Adam algorithm com- bines the advantages of Ada Grad and RMS Prop algorithms 4.2. Model Combinations. In order to fully utilize the ad- to dynamically adjust the learning rate of each parameter by vantages of the Prophet model and LSTM neural network using the ﬁrst-order moment estimation and second-order model,thispaperproposes anoptimal combinedforecasting moment estimation of the gradient. It is an eﬀective gra- model based on Prophet and LSTM neural network for sales dient-based stochastic optimization method, which makes time series forecasting of calligraphy exercise products. In the update of parameters smoother and takes less storage February 2019, Drotar´ et al. open-sourced a set of time series resources. forecasting tools Prophet. Unlike traditional time series forecasting models (ARIMA, etc.), the Prophet model is essentially a curve ﬁt to time series data, while having ex- 4.3. Experimental Results. In order to ﬁnd the optimal cellent adaptability to holiday eﬀects and trend change weight coeﬃcients w and w in the integrated model 1 2 points (change points) in the data, especially robust to Prophet-LSTM model, we take the coeﬃcients w to be 11 missing values, shifts in trends, and a large number of values in [0.0, 1.0] increments, w + w � 1 and then the 1 2 outliers, and is currently mainly applied to traﬃc forecasting corresponding weight coeﬃcients w to be 11 values in [1.0, on Facebook social networking sites. Prophet itself is a 0.0] decrements and multiply the 11 sets of weight coeﬃ- model based on a self-additive model to predict time series cients w and w with the sales volume forecasts of the 1 2 data, the model as a whole consists of growth (trend term), respective models at each moment. After the above process, seasonality (period term), and holidays (holiday term) 3 the11setsofweightedintegratedsalesforecastsareobtained parts superimposed, and the basic form is as follows: as P(t) � g(t) + s(t) + h(t) + ε, (5) Y t􏼁 , i � 1,2, . . . ,11, ti � 1,2, . . . , n. (7) Here, the trend term g(t) is the core component of the In order to ﬁnd the speciﬁc values of the weight coef- whole Prophet model, containing parameters with diﬀerent ﬁcients w and w , the root mean square error equation 1 2 degrees of assumptions and adjusting smoothness, which is (RMSE) and the mean absolute error (MAE) are used as the used to ﬁt the nonperiodic changes in the time series and criteriatoevaluatetheeﬀectoftheProphet-LSTMcombined select the change point (change point) from the data to sales forecasting model for each group of weight coeﬃcients, detectthetrenddirection.Firstly,theProphetmodelandthe andtheexpressions areas follows:wherexistheactualvalue LSTM network model with high prediction accuracy are of product sales in week i. According to Table 2, the root constructedseparatelyforthesalesdata,thepredictionvalue mean square error of the prediction results is minimized of the Prophet model at time t is set as P(t), the prediction when the weight coeﬃcients w �0.6 and w �0.4, and the 1 2 value of the LSTM network model is L(t), t �1, 2, ..., N, and mean absolute error of the prediction results is minimized the two models are assigned dynamic weights w and w , 1 2 when the weight coeﬃcients w � 0.4 and w � 0.6. *e 1 2 respectively. t, this point, deﬁnes the integrated Prophet- resultsshowthatthepredictionresultsoftheProphet-LSTM LSTM combined prediction model as model are improved compared with the Prophet and LSTM models, and its prediction accuracy is generally better than Y(t) � w P(t) + w L(t), w + w � 1, t � 1,2, . . . , N, 1 2 1 2 the other two single prediction models, indicating that the (6) prediction results of the Prophet-LSTM combined predic- wheretisthetimewhenthepredictedvalueappearsandY(t) tion model are relatively eﬀective and more applicable than is the result of summing the predicted data of the Prophet the single prediction model in the sales forecast of this model and LSTM network by weights. *e Prophet-LSTM enterprise. *e weight coeﬃcients corresponding to the neural network sales volume forecasting model is obtained evaluation metrics of the model are shown in Table 3. by calculating the optimal weights composed of the two In order to further verify the application performance of models after integration. *e Prophet model can transform the Prophet-LSTM model, Holt–Winters (exponential the time series by certain functions into combined patterns smoothing model), ARIMA model, Prophet model, LSTM, of diﬀerent time dimensions such as daily, weekly, quarterly, and Prophet-LSTM combined model were used as com- and yearly with overall trends. *e Prophet-LSTM model is parison models in this paper to model the prediction of ﬁrst initialized and empirically set change point �0.15 to another new set of sales data, and the evaluation results of make the growth trend more sensitive to changes. *e each model are listed in Table 3. *e evaluation results show Journal of Mathematics 9 Table 2: Comparison of the performance of diﬀerent models. Evaluation metrics Models RMSE MAE Holt–Winters 35.273 16.976 ARIMA 23.457 13.650 Prophet 3.465 2.143 LSTM 5.746 3.254 Prophet-LSTM (w �0.6, w �0.4) 2.645 1.645 1 2 Prophet-LSTM (w �0.4, w �0.6) 7.103 3.635 1 2 Table 3: *e weight coeﬃcients correspond to the evaluation metrics of the model. w 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.0 w 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 RMSE 3.700 3.427 3.274 3.153 3.059 3.022 3.010 3.050 3.139 3.253 3.490 MAE 2.514 2.380 2.300 2.274 2.269 2.297 2.351 2.453 2.524 2.605 2.778 that Prophet and LSTM single models have comparable forecasting model combining two single forecasting models forecasting performance, and both are signiﬁcantly better is designed, on which new forecasting models can be in- than Holt–Winters and ARIMA classical time series models; troduced for further combination in the future to pool the Prophet-LSTM combined forecasting model has further advantages of more models and thus obtain more accurate optimized the forecasting accuracy based on the single forecasting results. Besides, further experiments can be done model and has the best forecasting eﬀect. on the parameter preferences of LSTM neural networks in the follow-up work to seek a better modeling method. In the future, in order to analyze the sales volume time series data 5. Conclusion more deeply, further research on the inﬂuencing factors of sales volume can be targeted. Calligraphy is the art of writing Chinese characters. *e Chinese people love calligraphy and practice it. In a general sense, the art of calligraphy is the art of writing; writing can Data Availability onlymeet theaestheticrequirementsoftheartofcalligraphy *e data used to support the ﬁndings of this study are if it is pursued in an aesthetically pleasing way and subli- available from the corresponding author upon request. mated to art. From the practice of calligraphy education, learning calligraphy has many functions. *e market for calligraphy practice products is vast, but in the face of a Conflicts of Interest complex and dynamic market, it is important to predict the *e author declares that he has no conﬂicts of interest. market demand for calligraphy practice products in a timely and accurate manner, to analyze the characteristics and patternsofsalesvolumedata,andtoimprovetheaccuracyof References salesvolumeforecasting,sothatcalligraphypracticeproduct [1] S. Lin and X. Ban, “Study on cultural and creative design of enterprises can develop timely and eﬀective marketing traditional chinese calligraphy tools in digital era,” in Pro- strategies.*ispaperproposesacombinedmodelprediction ceedings of the 6th International Conference on Arts, Design method based on recursive neural network model based on and Contemporary Education (ICADCE 2020), pp. 427–430, the pattern of sales volume time series data of a calligraphy Jaipur, India, Febraury 2021. practice product enterprise, constructs a weighted Prophet- [2] A. Ahmad Zarnuji, H. Hanif Amrulloh, and I. N. Isnaini Nur LSTM combined prediction model, and conducts compar- Azizah, “Utilization of rice husk waste for paper raw materials ison experiments with the model under each weight coef- as an Arabic calligraphy media,” Engage: Jurnal Pengabdian ﬁcient, the single model before combination, and two Kepada Masyarakat, vol. 3, no. 1, pp. 43–54, 2019. classical time series models. *e experimental results show [3] Y. Dai, “Research on the application of computer multimedia in calligraphy education,” Journal of Physics: Conference Se- that the forecasting performance of the Prophet and LSTM ries, vol. 1915, no. 3, Article ID 032019, 2021. neuralnetworkmodelsissigniﬁcantlyhigherthanthatofthe [4] L. Xiang and L. L. S. Gee, “*e importance of lines: the typical time series forecasting models, and the combined modern calligraphy of wang dongling and qiu zhenzhong,” in Prophet-LSTM model has better forecasting performance. Proceedings of the 2nd International Conference on Interdis- Overall,basedontheveriﬁcationthattheProphetmodeland ciplinary Arts & Humanities (ICONARTIES) 2020, Yogya- the LSTM model have obvious advantages in forecasting the karta,IndonesiaAvailable atSSRN3800559,2021,Yogyakarta, sales volume of calligraphy practice products, this paper Indonesia, March 2021. further improves the model prediction performance by [5] R. Long, C. Sun, and H. Pan, “Research on the design and weighted combination, maximizing the advantages of both application of persuasive game in calligraphy learning,” in prediction models. In this paper, only a combined ProceedingsoftheInternationalConferenceonAppliedHuman 10 Journal of Mathematics Factors and Ergonomics, pp. 473–479, San Diego, CA, USA, July 2020. [6] Y.Li,“CalligraphyEducationinTeachingChineseasaSecond Language,” ;e Palgrave Handbook of Chinese Language Studies, vol. 5, pp. 1–29, 2020. [7] P. Lyu, X. Bai, C. Yao, H. Tengteng, and L. Wenyu, “Auto- encoder guided gan for Chinese calligraphy synthesis,” in Proceedings of the 2017 14th iapr international conference on document analysis and recognition (icdar), pp. 1095–1100, Kyoto, Japan, November 2017. [8] M. R. Rosenberger, C. K. Dass, H.-J. Chuang et al., “Quantum calligraphy: writing single-photon emitters in a two-dimen- sional materials platform,” ACS Nano, vol. 13, no. 1, pp. 904–912, 2019. [9] H. S. R. Kao, M. Xu, and T. T. Kao, “Calligraphy, psychology and the confucian literati personality,” Psychology & Devel- oping Societies, vol. 33, no. 1, pp. 54–72, 2021. [10] Y.-W. Zhang, Y. Chen, H. Liu, Z. Ji, and C. Zhang, “Modeling Chinese calligraphy reliefs from one image,” Computers & Graphics, vol. 70, pp. 300–306, 2018. [11] Y. Weng, X. Wang, J. Hua, H. Wang, M. Kang, and F.-Y. Wang, “Forecasting horticultural products price using ARIMAmodeland neuralnetworkbased on a large-scaledata set collected by web crawler,” IEEE Transactions on Com- putational Social Systems, vol. 6, no. 3, pp. 547–553, 2019. [12] A. Di Piazza, M. C. Di Piazza, G. La Tona, and M. Luna, “An artiﬁcial neural network-based forecasting model of energy- related time series for electrical grid management,” Mathe- matics and Computers in Simulation, vol. 184, pp. 294–305, [13] K. Muralitharan, R. Sakthivel, and R. Vishnuvarthan, “Neural network based optimization approach for energy demand prediction in smart grid,” Neurocomputing, vol. 273, pp. 199–208, 2018. [14] H. Eskandari, M. Imani, and M. P. Moghaddam, “Convolu- tional and recurrent neural network based model for short- term load forecasting,” Electric Power Systems Research, vol. 195, Article ID 107173, 2021. [15] I. Matino, S. Dettori, V. Colla, V. Weber, and S. Salame, “Forecasting blast furnace gas production and demand through echo state neural network-based models: pave the way to oﬀ-gas optimized management,” Applied Energy, vol. 253, Article ID 113578, 2019.

Journal

Journal of Mathematics – Hindawi Publishing Corporation

Published: Jan 5, 2022

References

Study on cultural and creative design of traditional chinese calligraphy tools in digital era,

S. Lin and X. Ban
Utilization of rice husk waste for paper raw materials as an Arabic calligraphy media,

A. Ahmad Zarnuji, H. Hanif Amrulloh, and I. N. Isnaini Nur Azizah
Research on the application of computer multimedia in calligraphy education,

Y. Dai
The importance of lines: the modern calligraphy of wang dongling and qiu zhenzhong,

L. Xiang and L. L. S. Gee
Research on the design and application of persuasive game in calligraphy learning,

R. Long, C. Sun, and H. Pan
Calligraphy Education in Teaching Chinese as a Second Language,

Y. Li
Auto-encoder guided gan for Chinese calligraphy synthesis,

P. Lyu, X. Bai, C. Yao, H. Tengteng, and L. Wenyu
Quantum calligraphy: writing single-photon emitters in a two-dimensional materials platform,

M. R. Rosenberger, C. K. Dass, H.-J. Chuang et al.
Calligraphy, psychology and the confucian literati personality,

H. S. R. Kao, M. Xu, and T. T. Kao
Modeling Chinese calligraphy reliefs from one image,

Y.-W. Zhang, Y. Chen, H. Liu, Z. Ji, and C. Zhang
Forecasting horticultural products price using ARIMA model and neural network based on a large-scale data set collected by web crawler,

Y. Weng, X. Wang, J. Hua, H. Wang, M. Kang, and F.-Y. Wang
An artificial neural network-based forecasting model of energy-related time series for electrical grid management,

A. Di Piazza, M. C. Di Piazza, G. La Tona, and M. Luna
Neural network based optimization approach for energy demand prediction in smart grid,

K. Muralitharan, R. Sakthivel, and R. Vishnuvarthan
Convolutional and recurrent neural network based model for short-term load forecasting,

H. Eskandari, M. Imani, and M. P. Moghaddam
Forecasting blast furnace gas production and demand through echo state neural network-based models: pave the way to off-gas optimized management,

I. Matino, S. Dettori, V. Colla, V. Weber, and S. Salame

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Recursive Neural Network-Based Market Demand Forecasting Algorithm for Calligraphy Practice Products

Recursive Neural Network-Based Market Demand Forecasting Algorithm for Calligraphy Practice Products

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Recursive Neural Network-Based Market Demand Forecasting Algorithm for Calligraphy Practice Products

Recursive Neural Network-Based Market Demand Forecasting Algorithm for Calligraphy Practice Products

References

Abstract

Journal

Recommended Articles

References

Our policy towards the use of cookies