摘要 |
<p><P>PROBLEM TO BE SOLVED: To provide an information processor which creates a classifier which discriminates whether a text document is the text document from which is likely to cause a result of incorrect correspondence analysis. <P>SOLUTION: A sample analysis section 4 performs correspondence analysis to a plurality of correct answer corpuses. A teacher data acquisition section 6 obtains, for each correct answer corpus, a result of determination whether or not accuracy of correspondence analysis performed to this correct answer corpus is beyond predetermined accuracy. A learning data acquisition section 8 obtains, for each correct answer corpus, a feature quantity corresponding to a reference learning item as learning data. A classifier creating section 10 creates a second classifier based on the teacher data and the learning data. <P>COPYRIGHT: (C)2010,JPO&INPIT</p> |