SENTENCE PERCEIVING DEVICE AND METHOD, PARTICULARLY CONCERNED WITH PERCEIVING A SENTENCE BASED ON A PROBABILITY MODEL BY USING DICTIONARY AND THESAURUS

摘要

PURPOSE: A sentence perceiving device and a method are provided to perceive a sentence based on a probability model by using a dictionary and a thesaurus, thereby enabling questions and answers as well as conversations with a machine. CONSTITUTION: A sentence recognizer(100) converts a voice signal and a key signal into a character string to output the character string. A word extractor(101) extracts words from the character string. A directional retrieval unit(102) reads explanations on a pre-selected category, compares the extracted words with the explanations, and calculates a probability value to output the value. A superordinate word retrieval unit(103) requests a thesaurus retrieval unit(104) to retrieve superordinate words, reads explanations on the superordinate words, compares the extracted words with the read explanations, and calculates a probability value to output the value. The thesaurus retrieval unit(104) reads and outputs the superordinate words from a thesaurus DB(110).