Summarizing Search Results using PLSI
Summarizing Search Results using PLSI
复制标题
DOI：
--
复制
发表时间：
2010-08
期刊：
影响因子：
--
通讯作者：
Jun Harashima;S. Kurohashi
中科院分区：
文献类型：
--
作者：
Jun Harashima;S. Kurohashi
在本文中，我们研究了从搜索结果中生成一组以查询为中心的摘要。由于在搜索结果中可能存在与给定查询相关的许多主题，为了汇总这些结果，应首先将其分类为主题，然后应单独汇总每个主题。在此汇总过程中，需要减少两种类型的冗余。首先，每个主题摘要不应该包含任何冗余（我们将此问题称为摘要中的冗余）。其次，主题摘要不应该与任何其他主题摘要相似（我们将此问题称为摘要之间的冗余）。在本文中，我们专注于文档聚类过程和减少摘要之间的冗余摘要过程。我们还提出了一种方法，使用PLSI总结搜索结果。评估结果表明，我们的方法在分类搜索结果和减少摘要之间的冗余表现良好。
In this paper, we investigate generating a set of query-focused summaries from search results. Since there may be many topics related to a given query in the search results, in order to summarize these results, they should first be classified into topics, and then each topic should be summarized individually. In this summarization process, two types of redundancies need to be reduced. First, each topic summary should not contain any redundancy (we refer to this problem as redundancy within a summary). Second, a topic summary should not be similar to any other topic summary (we refer to this problem as redundancy between summaries). In this paper, we focus on the document clustering process and the reduction of redundancy between summaries in the summarization process. We also propose a method using PLSI to summarize search results. Evaluation results confirm that our method performs well in classifying search results and reducing the redundancy between summaries.