Application Of TFIDF Keyword Extraction Algorithm Incorporating Validity Of Commentsin Electronic Product

With the rapid development of the Internet,China’s shopping market has shifted from offline to online.Potential consumers rely more on online-reviews when they can not directly perceive the products before buying them,so how to make the potential consumers quickly and efficiently capturing key information from the vast amount of online-reviews and helping businesses to understand the consumer’s need is critical,as well as the core of this study.This paper selects electronic products from many commodity categories as the research object.By analysising data characteristics,it is found that the traditional TFIDF algorithm,as a common method of keyword extraction,has some deficiencies in the application level of online comments and the theoretical level of weight calculation,also the huge amount of useless information in online reviews that affect consumers.To improve the performance of text analysis,this paper incorporates validity feedback from online-reviews based on the traditional TFIDF algorithm.The central idea is that if the more useful a review is,the more important the keywords extracted from that review are,the more weight is added to that review.Through empirical study and comparative experiments,it proves that the improved method works well,and then clustering analysis is carried out on the extraction of keywords.It is concluded that buyers concerns more information such as high price and quality,authentic,looks,sound,electricity,packaging and delivery logistics,finally corresponding suggestions are given from four aspects:manufacturers,shops,e-commerce platforms and logistics companies.