On Demand Recommendation Of IOS App Based On Weighted Word2vec Fusion Of Multidimensional Information

Under the background of big data and mobile computing,mobile applications at home and abroad of the mall APP emerge in endlessly,smart PC devices has become an indispensable part of people life and various apps arises at the historic moment,in the promotion of all kinds of APP,how accurate for users to recommend the suitable APP download is a challenging problem,has attracted wide attention of researchers both at home and abroad.However,most of the existing research methods recommend apps based on a single APP information,ignoring the multidimensional information that can reflect the popularity of apps.Therefore,this paper conducts an in-depth study on how to make high-quality APP recommendation based on multi-dimensional information of APP for IOS mobile users.The main research contents are as follows:(1)A similar APP recommendation method based on weighted Word2 vec is proposed.With the user requirements in the method for the APP Store and unrelated APP appear in the search list of problems,first using the crawler software APP of seven wheat data network information are collected,and the information collected by preprocessing,and then use TF-IDF(Term Frequency,Inverse Document Frequency,word Frequency,Inverse Document Frequency)algorithm from the APP description to extract key information,and the use of Word2 vec tools on information extraction of key APP description to quantitative.A similar method was used to conduct vectorization operation on the text of user requirements,and the cosine similarity calculation method was used to sort the similarity calculation.Top-N apps with high similarity to the text of user requirements were selected to get the recommendation list of similar apps.Finally,the real data on qimai data network are verified,and the experimental results show that the similar APP recommendation method based on Word2 vec proposed in this paper can effectively improve the recommendation accuracy and recall rate.(2)A Top-K APP recommendation method driven by multidimensional information is proposed.This method aims at the problem that the less popular apps in the APP Store search recommendation list appear at the Top of the list.Firstly,the LSTM(Long Short Term Memory Network)model is used to analyze the emotional polarity of pre-processed APP reviews,select the favorite apps,and obtain the candidate recommendation list of Top-K apps.On this basis,the popularity measurement model of APP was built according to the APP rating,the number of downloads,and the number of times the APP entered the APPStore for boutique recommendation,and the popularity measurement of each APP in the Top-K APP candidate recommendation list was conducted.Finally,the Top-K APP with a high degree of comprehensive similarity is recommended to users based on the comprehensive consideration of APP popularity measurement and the similarity between user search content and APP description.Through the experiment of real data on qimai data network,the experimental results show that the driven by multidimensional information Top-K APP recommendation method can effectively improve the NDCG value and MAP value of the recommendation results,and the validity of the method is verified.