摘要 |
PURPOSE: To provide a natural language processing system capable of automatically learning the case frame pattern of a sentence as a multidimensional probability model and using the result as semantic knowledge. CONSTITUTION: At the time of learning, an example extraction part 2 extracts an example from sentence data stored in a sentence data storage part 1 and a learning part 3 receives the example, refers to the hierarchy of a concept for each part of speech stored in a concept hierarchy storage part 4, uses an MDL standard, learns the semantic knowledge and stores the result in a semantic knowledge storage part 5. The semantic knowledge is expressed as case frames. At the time of a natural language processing, a natural language processing part 6 receives a natural language, refers to the semantic knowledge stored in the semantic knowledge storage part 5 and outputs a processing result. By referring to the semantic knowledge, the quality of the natural language processing is improved. |