Research On Fast Fashion Trend Based On Semantic Text Mining And Time Series Clustering ——Take The Trend Of Jeans As An Example

With the characteristics of low prices and quite changing styles,"fast fashion" clothing is one of the development trends of the clothing industry and always follows the trend of the season.As the mobile internet keeps expanding rapidly nowadays,how to grasp the characteristics of the rapidly growing Chinese fast fashion market and mine trends by using big data technology has become one of the strategic priorities of Chinese fast fashion enterprises at present.According to the public release of information in the social platform,through fast fashion-related trend features of jeans that are extracted from the post of jeans by text semantic mining technology,this paper would take jeans as an example to construct a fashion characteristic trend analysis model.Therefore,the characteristics of fashion trends and fashion features can be identified.In addition,this research can also provide scientific data support for fast fashion enterprises to formulate future clothing design strategies and realize application innovation from theoretical model research to practical problem-solving.Specifically,this paper selects "jeans",a representative single item of fast fashion,as the research object.Firstly,crawler technology is used to collect microblog data with "jeans" as the keyword from December 9,2021,to January 8,2022.The original data would be preprocessed and the Chinese word segmentation techniques would be used.Then,through constructing the semantic network of jeans fashion blog post data after text processing,10 core fashion feature words are extracted according to the results of co-occurrence analysis,semantic network graph analysis,and centrality analysis.Then,the LDA topic model of the jeans fashion blog would be constructed.According to the results of the confusion test and coherence test,the optimal number of topics would be determined.Then,the five core topics,each with 8 feature words,would be extracted.Finally,according to the text semantic mining results,combined with the actual situation analysis,22 hot fashion features of jeans are extracted as the research object of trend analysis.Furthermore,in order to explore the trend of jeans fashion features,this paper constructed a fashion feature trend mining model based on time series clustering algorithm and MannKendall test.By comparing the model effects of the four clustering algorithms,it is concluded that the k-Shape clustering algorithm with SBD as the distance measure and Shape extraction as the prototype function has the best effect on the fashion feature trend mining model.Finally,to further identify fashion features that are closely related to social attention changes,this paper proposes a same-trend fashion feature combination recognition model.The identification models before and after optimization will be established respectively according to the time series of the topic degree of fashion features and the time series of the normalized daily variation of topic degree.In addition,this paper also applies the recognition model to the results of the excavated fashion feature trends and verifies the improvement of the model effect before and after optimization.Based on the results of the two fashion trend analysis models,four fashion trends and their corresponding classification of fashion features were identified: "up","down","evergreen" and "down first and then up".Compared with the model before optimization,the SSE of clustering result in the optimized jeans fashion feature combination identification model is reduced by 34.69%on average,and the silhouette coefficient is increased by 16.25% on average.And the result of feature combination recognition has more application value.After that,the feature combination recognition model based on the optimization can identify 7 groups of feature combinations that are closely related to the change of topic degree,including "light blue stripe",and can be matched as a new design direction.Finally,according to the research results of this paper,combined with the star effect,seasonal effect and other factors,it puts forward objective and reasonable suggestions for fast fashion enterprises to make jeans design and sales strategy.