| 摘要 |
<p><P>PROBLEM TO BE SOLVED: To provide a knowledge construction device that generates and accumulates a rule for obtaining component information on a text, in a concept level, regardless of word surface layer. <P>SOLUTION: The knowledge construction device includes a text data storage unit which stores a text item, and component information in association with each other. A word extraction unit extracts a word specific to the text item, based on the appearance frequencies of words contained in the text items. A knowledge data acquisition unit acquires knowledge data showing a relationship between word and word concept. A concept acquisition unit uses the knowledge data to replace the word extracted from the text item with the corresponding word concept, combines it with component information associated with the text item, and outputs the result as case data. A mechanical learning device performs mechanical leaning by use of the case data, and generates a rule showing a relationship between the word concept and the component information. <P>COPYRIGHT: (C)2011,JPO&INPIT</p> |
