基于科技文献引用关系扩展的信念网络模型

信念网络信息检索模型提供了一个可以组合不同证据的基本框架,通过组合证据可有效提高检索效果.科技文献之间的引用关系是一种有价值的信息.本文把这种引用关系量化为文献关注度和文献价值度,并将其作为证据扩展基本信念网络模型.考虑到文档内容证据、关注度和价值度对查询结果排序的影响不同,引入了2个系数α和β来调整3种证据在排序中的作用大小.实验通过样本训练确定了α和β的最优值,验证了扩展模型的有效性,并分析了扩展模型用and和or方式组合证据的优劣.

Extended belief network model based on reference relationship of scientific literatures

The belief network retrieval model which provides a framework for combining different evidences is advantageous to improve retrieval effectiveness.The reference relationship between scientific literatures is a kind of valuable information.This reference relationship was quantified as attention and value in this paper,and it was the evidence for extending the belief network model.Consider the affections of the document content,attention and value were different for the sort of query results,we introduced two coefficients and to adjust the role of the three kinds of evidence for the sort.In our experiments,we determined the optimal value of and through certain sample training,testified the effectiveness of our extending model,and we also analyzed the advantages and disadvantages of the evidences combined by and and or.