CAJ | 학술논문

针对短文本单一共现词特征扩展效果不理想的情况,提出一种改进的基于共现关系的短文本特征扩展算法,改进之处在于考虑了多个共现词同时出现的情况,改进了特征词权重计算公式及特征扩展策略,并应用于中文短文本分类,使分类准确度得到了一定提升。
침대단문본단일공현사특정확전효과불이상적정황,제출일충개진적기우공현관계적단문본특정확전산법,개진지처재우고필료다개공현사동시출현적정황,개진료특정사권중계산공식급특정확전책략,병응용우중문단문본분류,사분류준학도득도료일정제승。
In this paper, an improved expansion algorithm based on co-occurrence relationship between short text feature is proposed aimed at not ideal situation for a single co-occurrence word feature expansion. The im- provement is that we considered more than a total of the current words at the same time and improved features of the word weight calculation formula and characteristics of expansion strategy, and applied to the Chinese short-text classification, which has improved classification accuracy.