Recommend User To Comment On The Information-based Goods

With the rapid development of World Wide Web (WWW), information grow rapidly and internet has become the important source that people obtained information. However, for the extention and opening of web, it is extremely easy and unrestricted to release information on internet, no matter as any unit, group or individual person. This has aggravated the information's inflation of WWW. So how we look for necessary information fast and accurate has been a difficult problem to users.At present, e-commerce grows vigorously at an incredible speed. In the new business environment, how to excavate out new potential users, and guarantee that one's own products can attract the customers, have become problems urgently to be solved of enterprise which launch e-commerce. Meanwhile, how the e-commerce website to find the tastes of users as much as possible, and recommend effectively, and adjust the production, propaganda work to satisfy users? So, how to offer the information for commercial website also has become the focus at present.Recommend technology on the basis of the internet information is exactly the effective means to solve the above difficult problems. It helps users to look for the necessary information from this information ocean of internet, and recommend the service people needed. It also can let the trade company get users'feedback information to its products at the same time . Recommended technology can facilitate users obtain the information on the network and accelerated the knowledge sharing. So, to carrying on the recommendation technical research not only has very high commercial value and academic value ,but also has important research meanings.This paper has done the work of several respects of the following mainly:1. User comment block's discovery and extraction: We put forward a kind of novel algorithm to find and extraction users'reviews in web pages.The new algorithm cut page apart into a lot of semantic piece first, then through calculate every semantic block's entropy value to extract the block of user review automatically. The experiment indicates that the new algorithm can get very high extraction quality and efficiency. 2. User review's emotion classification: We put forward a simple and effective method to complete the emotion classification. We make use of the synonym set and antonym set of the adjective in WordNet to orient a semantic of the adjective in reviews. The experiment show that the new algorithm can get the satisfied result of classification.3. Popular goods are recommend technology: We analys the users'reviews and recommend products for users based on them finally. Through analysing the semantic information of every review, we can count good/bad comment of total amount which a certain goods receive. Because when users use search engine to search for goods, they all hope that it can return the pages of products which are most popular and have best quality and value . So, we proposed a kind of new page ranks algorithms to rank product pages. The algorithm combines product page own importance and this product's favorable comment rate, then carry on re_rank the pages that search engine return. The experiments indicate that the new algorithm can improve the users'satisfaction of the result to product pages which search engine return.