Research On Recommendation Method Based On Text Analysis

With the advent of the information age,the Internet is flooded with more and more data information,making it impossible for users to get part of the data information of interest when facing a large amount of data information,causing users to waste time and energy,and the competitiveness of the network platform decline.Net Ease cloud music users face a huge amount of music songs,how to choose a user is a problem,the recommended method can help users find the songs that users like.The previous recommendation method is for the classification or scoring of music,and does not take into account the user's evaluation of it.Therefore,this article uses the user's review text as data to conduct research on the recommendation method.Different from the existing recommendation methods,the text of Net Ease Cloud Music user reviews is used as the data object,and the text statistical analysis method is adopted for the user review texts,including Chinese text jieba word segmentation,TFIDF algorithm,and k-means clustering.The first is to crawl the comment text of Netease Cloud Music with python and nodejs,and save it as a.csv file.The differences and characteristics of Chinese and English texts are analyzed,and then the Chinese text segmentation is completed using jieba word segmentation.The word segmentation is combined with a dictionary plus stop words and statistical methods to make the word segmentation more reasonable.The concept and idea of the TF-IDF model are analyzed,and the TF-IDF calculation is performed on the word segmentation,so as to transform the unstructured text data into structured vector data.Perform k-means clustering analysis on TF-IDF.According to the classification results obtained by k-means clustering,draw a word cloud named theme,and recommend theme songs for classified users.