| 摘要 |
<P>PROBLEM TO BE SOLVED: To allow a result of learning processing based on learning data to be used for extracting a sentence section from a plurality of sentences and to shorten a learning processing time. <P>SOLUTION: The text analysis device is provided with a final hypothesis information storage part storing final hypothesis information, a first tree structure generation part generating a tree structure based on a result of syntax analysis of the learning data, a weak learner generation part extracting a plurality of subtrees from the generated tree structure for generating weak learners, a learning processing part performing learning processing by Gibbs-Boost learning algorithm based on a plurality of generated weak learners for generating final hypothesis information, a second tree structure generation part generating a tree structure corresponding to a sentence section included in text data from the text data to be determined, and a determination processing part determining whether the sentence section corresponds to a retrieval target or not based on the tree structure and the final hypothesis information. <P>COPYRIGHT: (C)2008,JPO&INPIT |
