QUESTION AND ANSWER SYSTEM WHICH CAN PROVIDE DESCRIPTIVE ANSWER USING WWW AS SOURCE OF INFORMATION

摘要

Provided is a technology to receive a question expressed by ordinary words, extract candidate answers for the question from among the documents on the WWW, and provide the answers to the user, which reflects formal correlations between questions and answers. A query sentence (1001) is converted into strings of words (1002) for formalizing the query sentence by conducting formalization processing wherein functional words, interrogative words, or predetermined words such as “reasons”, “meanings”, etc. which tend to be asked are converted into surface expressions and the other content words are converted into word class types. From among the strings of words (1002), key parts (1003) for a query sentence format, each consisting of an interrogative word and a predetermined number of strings of words before and after the interrogative word are extracted. Reference query sentences in question and answer samples are converted in like manner. Then, the aptitude of the format is judged from a degree of similarity of key parts (1004) for a reference query sentence format. Furthermore, reference answer sentences in the question and answer samples are also formalized and a degree of formal correlation with the query sentence is determined.