Integration of Automated Summarization and Question Answering in Interface for Information Retrieval

近年来，大量的文件可供我们使用的搜索引擎在互联网上。然而，我们不得不阅读大量的文本，以获得必要的信息，因为搜索引擎通常返回原始文档，仍然需要检查它们。因此，自动摘要和问题回答作为减少要阅读的文本量的技术而受到关注。在本研究中，我们的目的是减少用户的负担，通过引入自动摘要和问题回答到用户界面的信息检索。本研究的主要贡献归纳如下。（1）提出了一种基于A^* 搜索的实时问答系统及其改进。基于NTCIR4 QAC2，一个处理日语问答的评估研讨会的性能评估。（2）建议一个交互式图形用户界面，该界面具有自动向用户推荐某些相关文件和显示检索到的文件摘要的功能。（3）提出一种回答一系列相关问题的方法。（4）提出一种方法来回答那些要求对某物进行定义的问题。

Recently, a large number of documents are available to us by using the search engines in the Internet. However, we have to read a large amount of text in order to obtain necessary information because the search engines usually return the original document and it is still necessary to examine them. Therefore, the automated summarization and the question answering have gained attention as technologies to reduce the amount of text to be read. In this study, we aimed at reducing the load on users by introducing the automated summarization and the question answering into the user interface for information retrieval. The main contribution of our study is summarized as follows.(1)Proposal of a real-time question answering system based on A^* search and its improvement. The evaluation of its performance based with NTCIR4 QAC2,an evaluation workshop dealing with Japanese question answering.(2)Proposal of an interactive graphical user interface that have the function to automatically recommend some relevant documents to users and that to display the summaries of retrieved documents.(3)Proposal of a method to answer a series of related questions.(4)Proposal of a method to answer questions which ask definitions of something.