The referential properties of noun phrases in the Japanese language, whichhas no articles, are useful for article generation in Japanese-English machinetranslation and for anaphora resolution in Japanese noun phrases. They aregenerally classified as generic noun phrases, definite noun phrases, andindefinite noun phrases. In the previous work, referential properties wereestimated by developing rules that used clue words. If two or more rules werein conflict with each other, the category having the maximum total score givenby the rules was selected as the desired category. The score given by each rulewas established by hand, so the manpower cost was high. In this work, weautomatically adjusted these scores by using a machine-learning method andsucceeded in reducing the amount of manpower needed to adjust these scores.
展开▼