Research Of Retrieval Ranking Algorithm Oriented To Blog

Along with the increasing effect of blogs,They are used by more people in mang fields to express thoughts and to exchange,constituting a blog sphere with the form of a website.When blog grows rapidly,it is urgent to explore blog search.Although the technique of existing web search is mature, it is not suitable for blog retrieval ranking,because a blog is different from a traditional website. About how to find important blog-posts or blogs from numerous blogs,the most important questions are how to define their importance and how to rank them according to the importance. A blog consists of a series of blog-posts that are the targets of a seeker. Therefore the blog-post can be an evidence of blog ranking.In fact,there are many ranking factors such as content links and date.The first two are selected to analyze the importance of blog-posts.The paper proposes a SGRM (Space Graph Ranking Model) algorithm to rerank the blog-posts according to the BE(Blog Evidence).First,the concept of BE is defined in the triple that integrates content information of the blog-posts and blogs relation information represented by a matrix separately. The relation matrix of "post-blog" is denoted by the transition from post-post to blog-blog because of the defect of the sparse of links.The content matrix of "post-term" is denoted by typical VSM(Vector Space Model).Second,PCA(principal component analysis) is used to reduce their dimensiones,and CG(conjugate gradient)is used to combine the dimension-reduced matrixes,a LSS(Latent Semantic Space)matrix computed. KNN (K Near Neighbour) is used to construct the LSG of LSS.Finally,a regulation ranking framework is proposed to rerank the initial blog-posts got by the given query based on the reranking thought and improved LE(Laplacian Eigenmaps).The ranking algorithm SGRM of blog is concluded.We implement the following three groups of experiments. First, the best parameter of the ranking framework is got by comparison in different settings.Second,we compare the SGRM with other algorithms,and the front shows better performance. Finally we analyze the effect of the construction of LSG on blog ranking.