This paper presents a keyword extraction method by first calculating word weight with ATF×PDF(Average Term Frequency *Proportional Document Frequency) and then determining the keywords by ajoint weighconsidering the semantic si milarity between words.

This paper presented a keyword extraction in multi-document based on TextRank method,calculated words weight with ATF~*PDF for selecting candidates,and constructed TextRank model rely on semantic relation between candidates,iterated the graph-based ranking algorithm until convergence,at last generated a list of keyword and extract- ed keyword.

The method of TF*IDF can only be used for calculating words weight of single document,the paper proposes a method that is named as ATF*PDF (Average Term Frequency*Proportional Document Frequency) for calculating words weight of multi-document First calculate word s weight and select candidates,and then combine with semantic similarity between words so as to extract keywords.