A Study Of Personalized Search Based On Blog Content

One hundred people, one hundred needs. However, almost all the existing search engines take model "one search, all people" for granted, only leaving people drown in the excess and low effect information. Their backgrounds, interests, hobbies, behavior and retrieval environments are totally ignored. Although the third generation of search engines has not come yet, it is a trend that personalization is undoubtedly to solve these problems. Personalized Search is very simple, which means that the more search engines know about you, the more irrelevant search results it would eliminate for you. As a result, personalized search engines is no longer a software tool inhospitable just for retrievaling useful information, but a friend who helps us understand the world and ourselves much better.In order to achieve the goal of personalized searching, this paper decide to dig people's blog articles to gain the user's interests, in which information people are willing to go public. In this way, we succeed in protecting people's privacy, which usually be harmed in other way gaining user interests, such as click stream and the user's internet history. And With the help of open source web spider and search library, a personalized search engine based on blog contents has been implemented. It has two main features: personalized search ranking and personalized recommendation.In the subsystem of personalized search, we redefine the similarity formula between query vector and document vector, combining the factor of user interests, on the basis of analyzing the traditional ranking algorithms. Traditional search engines only build the relationship between keywords and documents. While here we succeed in building relationships among keywords, user and documents by using the vector space model and improve the search results greatly.Personalized recommendation subsystem implements the function of blog articles recommendation and blog user interaction by calculating the similarity of interests among users. Finally, like-minded people are able to get to know each other by blog user recommendation subsystem.In the last part of this paper, we briefly analyze some problems that this system might face in the future, put forward two possible development directions and draw a bright future for the perfect search.