Research On Vertical Search Engine Of Mobile Products Information

With the development of Internet technology, the flourish ofe-commerce and the rise of the Forum, blog, etc., more and more peoplelike to give their comments to the purchased goods, to express theirattitude, ideas and the use of feelings. Therefore, there are more and moreproduct reviews information on the network. By reading these reviews,the potential buyers can understand the characteristics of the product andmaking decisions whether to buy，besides, businesses can understand thesupply and demand relationship, popularity of goods timely and effectiveso that they can make their own decisions. But only relies on artificial tobrowse this information is time-consuming, and the information obtainedis one-sided, time-consuming, inefficiency.therefore, people increasinglyrely on search engines when searching for information. However, thegeneral search engine’s shortcoming is obvious in the specific areas, so itis very necessary to build a product-specific vertical search engines.Through analysis the vertical search engine’s and sentimentclassification’s research status at home and abroad，build phone productinformation vertical search engine as clues, the main work is as follows:(1) Design theme crawler framework for the field of mobile phoneproducts, on the basis of deep research the traditional content-basedsearch strategy and link search strategy, a combined search strategy basedon content and link was proposed. Greatly increased the relevance of thesubject of the crawling web page, facilitate the follow-up steps to build avertical search engine. Experiment comparing with the hits algorithm,breadth-first algorithm and pagerank algorithm shows the advantages ofthe proposed algorithm.(2) After obtain the phone product attributes and emotional words, acollocation identification method of attribute words and emotional wordswas proposed，use SVM method to train classifier, effective access theemotional tendencies score of review,then composite all the comments ofsome mobile model to compute the overall satisfaction. Verify theeffectiveness of the collocation method by experimental comparison. (3) Design and implementation a simple search engine for mobilephones product information, presents the design framework, and describethe implementation of each module，show the system interface.