| 摘要 |
PROBLEM TO BE SOLVED: To automatically register an unknown word onto a dictionary while estimating its part of speech by extracting the unknown word by writing an inputted Japanese character string while dividing it into clauses, and discriminating the part of speech of the unknown word from the constitution or attribute of connected words before and after the unknown word. SOLUTION: A morpheme analytic part 50 refers to the noun information or the like of respective words registered on an independent word dictionary 62 or adjunctive word dictionary 64, writes the inputted Japanese character string while dividing it into clauses (performs morpheme analytic processing), sends the result to a morpheme output part 52, outputs it through an input control part 46 to an output part 42 or display part 42 and sends the analyzed result to an unknown word extracting part 56. The unknown word extracting part 56 extracts the unknown word, which is not registered on the dictionaries 62 and 64, from the dividedly written clauses and dispatches it to an unknown word registering part 60. The unknown word registering part 60 discriminates the part of speech of the unknown word based on the constitution of connected word or attribute data in the language of connected words stored in the dictionaries 62 and 64 while extracting the connected words before and after the unknown word from the character string and registers it on the dictionaries 62 and 64. |
