摘要 |
<p>Disclosed is a method of analyzing documents, which employs a device 1 for analyzing documents that treats a collection of documents, comprising a time information, as an object of analysis. The device 1 for analyzing documents comprises a variance estimation unit (20), which estimates variances in a collection of documents (1) based on an information, i.e., a collection of documents (2), other than the collection of documents (1), a mining unit (30), which mines the text of the collection of documents (1) and detects characteristic words, using the estimated variances as criteria, and within a set interval, and an analysis unit (40), which acquires a time series data of a document that contains the detected characteristic words, and identifies the characteristic of the collection of documents (1) for the time series data thus obtained, based on the change of the characteristics before and after, wherein the variances are treated as the criteria thereof.</p> |
申请人 |
NEC CORPORATION;NAKAZAWA, SATOSHI;ANDO, SHINICHI;KAWAI, TAKAO;ONISHI, TAKASHI |
发明人 |
NAKAZAWA, SATOSHI;ANDO, SHINICHI;KAWAI, TAKAO;ONISHI, TAKASHI |