基于主题向量模型的多样化资源选择算法

Topic vector model based resource selection algorithm for diversification

检索结果多样化任务下的资源选择是联邦检索的重要研究内容。针对现有显式方法的信息依赖和隐式方法的多样化性能不高,提出在α相关度筛选后得到的相关文档集上,构建子主题向量模型。从文中语义层提取数据源内容特征,提升资源相似度计算准确度,实现多样化的资源选择。实验显示,基于主题向量模型的资源选择方法表现出更好的性能。

Resource selection under the diversified tasks of search results is an important research topic in federal search. The information-dependent for explicit methods and poor performance for the existing implicit methods,a sub-topic vector model is proposed based on the relevant documents obtained after theα-correlation filtering. It extracts the content characteristics of data sources from the textual semantic layer,enhances the accuracy of resource similarity calculation and diversifying resource selection. The experiments show that the resource selection method based on the topic vector model has better performance.