An Improved PageRank Algorithm Based on Latent Semantic Model
Xiaoyun Chen, Baojun Gao, Ping Wen · 2009
The traditional PageRank (PR) just takes into account the Web link structure, when distributing rank scores it treats all links equally, which results in topic drift. In this paper, latent semantic model (LSM) is used to calculate the similarity between Web pages, and the LSMPageRank (LPR) algorithm is introduced. In this algorithm, the value of parent page is distributed to the child on the basis of page similarity between them. The experiment which combines with Nutch shows that the LSMPageRank algorithm performs better than the PageRank algorithm and retrieves better result set.