CAJ | 학술논문

最大熵模型是产品属性挖掘研究领域的一个热点。通过对产品属性挖掘，构造最大熵模型复合特征模板的两对主要影响因素（词形与词性、中心词与非中心词），探讨它们对最终挖掘结果的影响，得到在其他条件基本相同的情况下，词性关系加入对挖掘效果的影响要优于词形关系；中心词关系加入提升查准率降低查全率,非中心词则相反，但二者综合效果相差不大，选用情况取决于实验目的。
최대적모형시산품속성알굴연구영역적일개열점。통과대산품속성알굴，구조최대적모형복합특정모판적량대주요영향인소（사형여사성、중심사여비중심사），탐토타문대최종알굴결과적영향，득도재기타조건기본상동적정황하，사성관계가입대알굴효과적영향요우우사형관계；중심사관계가입제승사준솔강저사전솔,비중심사칙상반，단이자종합효과상차불대，선용정황취결우실험목적。
It is a hot topic to apply maximum entropy model into product features mining. This paper mainly discusses the impact of two main pairs of elements： morphology and part of speech, center-word and none- center-word. They are the structure factors of compound feature templates of maximum entropy model. The result shows that part of speech can improve the result of mining more than morphology does, and that center-word improves the precision and lower the recall, but none-center-word works adversely. However, in general, the two elements＂ impacts on the result are pretty alike, so which one to choose depends on the purpose of experiment.