Keyphrase extraction based on topic feature
基于主题特征的关键词抽取

Keyphrase extraction is a process for extracting a set of terms from a document. This paper proposed a novel topic feature for extracting keyphrase. This topic feature was computed based on topic model which modeled the topic-word distributions and the topic distributions of document. Moreover, it proposed a keyphrase extraction approach based on bagged decision trees. This approach jointed common features and the proposed topic feature. Experimental results demonstrate that the proposed topic feature can make an improvement for keyphrase extraction. At the mean time, an effective performance can be achieved by the bagged decision trees based approach.