Study On The Personalized Appreal Recommendation Based On Hierarchical Vector Space Model

The rapid development of e-commerce provide users with more and more services and products,but also appeared the problem of information overload.With economic and cultural development,consumers no longer meet the basic functions of clothing.Clothing and dress began to pursue the individuality and taste.Recommended technology has great potential in resolving information overload and personalized recommendation,and has been applied to many fields.However,due to the influence factors of clothing preference are more and more unstable,and there is obvious difference psychology,the proposed technology proposes a higher Claim.Therefore,the study of personal recommendation for clothing is particularly important.From the Internet platform point of view,this article based on the hierarchical vector space model to build clothing personalized recommendation model.The main contents of the model are as follows:(1)Comprehensive literature research,consumer depth interviews and expert opinion are three ways to get the minimum attribute set that can describe the garment in a comprehensive way.The collection of garments is constructed based on the attribute of the garment,and the garment is represented by the vector space model.(2)According to the user's personal information and behavioral data,this article reference to the established clothing model,extract user attributes and classification.Based on the hierarchical vector space model,a user model is established,and the user model is divided into the user requirement model and the user preference model according to the influence of time and scene changes on the drift of interest.User demand model reflects the user's short-term needs,the main target user recommended clothing screening;user interest feature vector representation of user preferences model reflect the user preferences for different clothing attributes.(3)The hybrid recommendation algorithm is used to match the clothing model with the user model.First,the target clothing collection is obtained based on the screening rules,and then the overall score of each clothing in the target clothing collection is calculated according to the preference of the consumers.This article based on the TOPN recommendation list Form to complete the recommendation.In order to validate the effectiveness of the proposed personalized clothing model,80 female college students as experimenters to simulate the experiment.First,this article use web data collection software Archer to collect Taobao clothing,through data preprocessing to select 5000 representative costumes to build clothing collection;Second,access to the experimenter Taobao clothing historical purchase records and recent browsing history,enter the recommended model to get the recommended results;Finally,the model obtained recommended results and random recommended results shuffling sequence recommended to the experimenter to score.Based on the ratings of the recommended products based on experimenters' preference and purchase intention,the recommendation system was evaluated from two evaluation indexes:forecast accuracy and ranking accuracy.The results show that the proposed model proposed in this paper has a significantly higher prediction accuracy than the random recommendation,and the average Spearman coefficient is about 0.88,which proves that the proposed personalized model is effective.