Research Of Blog Posts Reranking Technology Oriented On User Intention

With the rapid development of Web2.0, blog begins to be concerned and used by more and more people as one of the typical applications. Being different from the common web page, blog post contains more abundant artificially tagging information which can edit by blogger anytime. It makes blog post very personalized. The feature that blog can be managed freely by user and the rapidly increasing of blogs bring the research of blog information processing a big challenge. How to help user find pages that meet his interest rapidly in so many blog posts becomes a hot research currently. Though existing related research of blog search has begun to focus on the characteristics of blog on ranking blog posts, rarely solve the problem from user’s perspective, and the result of reranking can not always satisfy the user, so it brought the research of blog posts reranking technology oriented on user intention.The blog posts reranking methods oriented on user intention in this thesis focus on behavior of user and characteristics of blog post, guide user to express his search target in a particular way, and return the result set according to the explicit user intention to user iteratively. Firstly, this thesis analyzes the major semantic attributes of blog post, and defines a model of blog posts reranking oriented on user intention on that basis and then gives an overall description of the process and framework of reranking. Next, the key algorithms in process of reranking are described in particular. The target of the blog posts clustering algorithm based on semantic extension is to partition all related results that meet the keywords into different classifications that each covers a subject. In the description of the algorithm, the process of clustering is proposed and each detail in process is described formally, the experimental results and conclusion based on this algorithm are given at last. The target of the blog posts sample selection algorithm oriented on user intention based on clustering is to select a certain number of blog posts that meet user interest from those classifications partitioned before. The algorithm is proposed according to some principles and the greedy process is carried according to constructing the sample space based on some heuristic rules first and selecting blog posts based on a special measurement function next. At last, this thesis introduces application and test detail of the technology of blog posts reranking oriented on user intention and thus proves the effectiveness of the algorithm that this thesis proposes.