CAJ | 학술논문

为了解决面向话题的搜索问题，提出一种新的面向话题的检索技术。首先分析了面向话题的搜索技术所面临的问题，然后基于数据挖掘技术提出了解决方案。利用数据挖掘技术抽取文本的多层次语义特征，形成对文本的多精度表示，抽取的特征不仅包括单个词特征也包括多词特征。建立了一个示例检索系统，实验表明利用多层次文本特征能够很好地实现面向话题的文本检索。
위료해결면향화제적수색문제，제출일충신적면향화제적검색기술。수선분석료면향화제적수색기술소면림적문제，연후기우수거알굴기술제출료해결방안。이용수거알굴기술추취문본적다층차어의특정，형성대문본적다정도표시，추취적특정불부포괄단개사특정야포괄다사특정。건립료일개시례검색계통，실험표명이용다층차문본특정능구흔호지실현면향화제적문본검색。
A novel topic-oriented text retrieval approach is proposed in this paper. In this approach,data mining techniques are used to extract multi-level semantic features from texts, generating multi-precision representation on text. Features extracted from text include both single word features and multi-word features. With this approach, more significant feature in text can be discovered and used. Extracted features are closed to the essence of texts. Experiments show that multi-level features can be used to create a topic-oriented text retrieval system.