摘要
提出词分布分析的方法,通过词的分段频数离散度和出现位置四分位数计算出查询词的分布权重,并将其与已有的基于向量空间模型(VSM)的算法和BM25算法相结合,改进了已有排序算法。实验结果表明,该方法对已有算法的排序效果确有提高。
A term distribution analysis approach is proposed. The distribution＇s weight is calculated through term＇s segmented frequency dis- persion and term＇s occurrence position＇s quartile,combined with the currently existed VSM based algorithm and BM25 algorithm, which im- proved the existed ranking algorithms. Experimental results show that the existed algorithms＇ ranking results are truly improved by this new approach.
出处
《世界科技研究与发展》
CSCD
2013年第1期49-51,108,共4页
World Sci-Tech R&D
关键词
词分布分析
相关排序
权重计算
信息检索
term distribution analysis
relevance ranking
weighing
information retrieval
