| 摘要 |
<P>PROBLEM TO BE SOLVED: To provide a keyword extracting device accurately extracting a keyword indicating a characteristic of an object document by properly using an evaluation item of another character string. <P>SOLUTION: Object document data is acquired by an input part 10, a document format is determined from the object document data on the basis of components in the document, and layout information, font size information, and appearance frequency information are generated from the object document data. A condition part describes an evaluation item state of evaluation item elements of a position, a font size, and an appearance frequency of a morpheme of the object document, and the consequent part describes whether it is a keyword or not. Since the layout information, the font size information, and the appearance frequency information of the object document are inputted in a working memory of a production stem 27 having a knowledge for each document format, and the production stem 27 carries out reasoning, by determining the document format of the object document, and using the layout information, the font size information, and the appearance frequency information of the object document. Only the production rule of the object document can be properly selected, and the keyword can be extracted by carrying out exact reasoning. <P>COPYRIGHT: (C)2007,JPO&INPIT |
