| 摘要 |
PROBLEM TO BE SOLVED: To provide high-speed and high-accuracy morpheme analysis in comparison with conventional analysis. SOLUTION: When there is a morpheme string having at least two constitutive morphemes among morpheme string retrieved at respective character positions (Y in step S16), analysis is performed only for the constitutive morphemes of that morpheme string having at least two constitute morphemes and analysis is not performed for other morpheme strings (steps S17-S22). Thus, when the morpheme string having at least two constitutive morphemes is retrieved, analysis is simplified (skipped) so that a processing speed can be accelerated. Especially, by using the morpheme string of a variable length in comparison with conventional longest match principle using a single morpheme, long- distance skipping is enabled and the processing speed is accelerated. Further, analysis based on matching over a wide range is enabled and lowering in analytic accuracy can be prevented. |
