摘要 : We describe an experiment mapping semantic role preferences for transitive verbs to their deverbal nominal forms. The preferences are learned by data mining large parsed corpora. Preferences are modeled for deverbal/argument pairs... 展开 We describe an experiment mapping semantic role preferences for transitive verbs to their deverbal nominal forms. The preferences are learned by data mining large parsed corpora. Preferences are modeled for deverbal/argument pairs, falling back to a model for the deverbal only when sufficient data is not available. Errors in role assignment are reduced by 30–40%. 收起