The traditional PageRank algorithm evaluates the importance of web pages through link relations between web pages and then ranks web pages. Because it only considers link relations between web pages, and ignores the relevance of the subject between web pages, web pages ranking quality is not high. In this paper, by full playing the HowNet characteristics of detailed describing semantic knowledge and revealing the relationship between concepts, I improved PageRank algorithm. The improved algorithm allocates PageRank value according to the semantic similarity of web page keywords in HowNet taxonomy tree. Through simulation experiments, the conclusion can be drawn that the new algorithm ranks more accurately than the traditional PageRank algorithm.
展开▼