| 摘要 |
<p>PROBLEM TO BE SOLVED: To correctly recognize the contents that notation having a large variation like the model name of a product, a sound-effect word, a numeral expression, etc., included in a document. SOLUTION: The natural language processor converts word notation expression grammar description data 101 wherein the constitution rule of a word belonging to a category having a large variety of notation is described to word notation context free grammar data 102 represented in the form of extended context free grammar. An analyzing procedure based upon the context free grammar is followed to cut a character string satisfying the constitution rule of the model name of a product, a sound-effect word, a numeral expression, etc., as a word out of an inputted natural language sentence according to the word notation context free grammar data 103. Further, the part of speech, reading, etc., of the word are determined by referring to a notation-unspecified dictionary 104 wherein word information is described in category units and the inputted natural language sentence is outputted in voice.</p> |
