APPARATUS AND METHOD FOR PARAGRAPH SEGMENTATION, AND SEARCH METHOD USING THE PARAGRAPH SEGMENTATION

摘要

The search method using the short-splitting device and method, and short-circuit division method is disclosed. Subjects setting section sets at least one or more subjects in the training data representing the predetermined domain. Pattern extracting unit divides the training data as a short message through the language analysis and extracting a first pattern consisting of a predetermined vocabulary from a pair of short divided respectively. Pattern learning unit to generate a thematic pattern DB classified according to the set of the extracted first pattern subject. Paragraph divider input document by the division into short by the language analysis and extracts the subject of the extracted a second pattern consisting of a predetermined vocabulary pair from the short of the divided each most similar pattern and the second pattern from the theme pattern DB short the allocation of the subject. Thus, the e-mail or an encyclopedia article, the search efficiency as well as easy to understand the short-star subject of a specific domain by offering to select only the subject you want in the application system, including questions and answers, information retrieval, such as newspaper articles It can be improved.