摘要 |
PURPOSE:To extract the meaning of an input voice as a natural language by outputting meaning relation defined by the status transition of an automaton generated by respective words in a word string as the meaning relation of the input voice. CONSTITUTION:A meaning relation storage part 103 previously stores words in a word string and the meaning relation of the words as the meaning relation of status transition strings corresponding to respective words when the word string is received by the automaton. A meaning relation extracting part 106 checks whether two groups of status transition strings out of the status transition strings received from a voice recognition part 105 are respectively stored in the storage part 103 or not, and when the data are stored, extracts their meaning relation and outputs the relation as the meaning relation of words generating the status transition. Consequently, the meaning of an input voice can be extracted as the meaing of a natural language without executing the syntax analysis and meansing analysis of the voice recognized result of the input voice. |