Abstract: Quantification obtained by quantifying the degree of similarity among synonyms by term similarity accurately, is used to improve the simple Bayesian network for information retrieval and for carrying out an effective probability inference. Experimental results show not only a good retrieval effectiveness of the new model, but a more reasonable ranking of relevant documents.

Key words: Bayesian networks, synonyms, information retrieval, similarity of term

摘要： 利用术语相似度将同义词间的相似程度数量化，以此量化关系对用于信息检索的简单贝叶斯网络进行改进，并进行有效的概率推理。实验结果表明新模型不仅具有良好的检索效果，而且相关文档的排序更加合理。

关键词: 贝叶斯网络, 同义词, 信息检索, 术语相似度

CLC Number: 

• TP391