CAJ | 학술논문

文本分类是解决网络信息过载的关键技术之一.传统的文本分类方法大多只考虑文本中词语的统计词频等特征,忽略了文本的语义信息,导致文本分类精度不高.针对这种问题,提出了一种基于语义距离的文本分类方法,该方法首先根据 CHI 方法进行文本特征选择,然后利用语义距离计算代表类别的特征向量集合,最后通过计算文本特征向量和类别特征向量之间的语义距离来确定文本类别.实验结果表明,该方法与其他方法相比,把文本的语义信息考虑在内,在进行文本分类方面具有较高的准确率.
문본분류시해결망락신식과재적관건기술지일.전통적문본분류방법대다지고필문본중사어적통계사빈등특정,홀략료문본적어의신식,도치문본분류정도불고.침대저충문제,제출료일충기우어의거리적문본분류방법,해방법수선근거 CHI 방법진행문본특정선택,연후이용어의거리계산대표유별적특정향량집합,최후통과계산문본특정향량화유별특정향량지간적어의거리래학정문본유별.실험결과표명,해방법여기타방법상비,파문본적어의신식고필재내,재진행문본분류방면구유교고적준학솔.
Text classification is one of the key technologier solving network information overload. The traditional text classification meth-ods only consider mostly the statistical word frequency in the text,ignoring the text semantic information,leading to text classification ac-curacy is not high. Aiming to this problem,propose a text classification method based on semantic distance,this method first take the text feature selection according to CHI method,and then by using semantic distance calculate feature vector set on behalf of the category,final-ly through the calculation of the distance between text feature vector and category feature vector determine the text category. The experi-mental results show that this method,compared with the existing methods,take the semantic information of the text into consideration,has higher accuracy in the text classification.