期刊名称：Lecture Notes in Engineering and Computer Science
印刷版ISSN：2078-0958
电子版ISSN：2078-0966
出版年度：2018
卷号：2237&2238
页码：308-313
出版社：Newswood and International Association of Engineers
摘要：The aim of the study is to analyze the semantic
relationships among utterances in a meeting at a deep level.
To achieve the aim, we propose a novel tree structure, called
discussion time-span tree, for representing a discussion structure
based on the relative importance of utterances contained
which are derived from the meaning of a meeting. A discussion
time-span tree is a binary tree, each leaf of which corresponds
to a utterance; it has the following characteristics: adjacent
relevant utterances are hierarchically grouped, and a more
important utterance becomes a head. Thus, a discussion timespan
tree represents implicit meaning and intention contained
in meeting records, and can be used for summarization. In the
study, we propose a method and rules to generate a discussion
time-span tree based on verbal and non-verbal information.
In the evaluation on important utterance extraction using the
proposed system, it was shown that the score of ROUGE-2
was higher than that of the existing document summarizing
technique. Moreover, in the evaluation by subjects conducted
by 10 subjects, we showed that the system can grasp conference
contents efficiently, and demonstrated the usefulness of the
system.
关键词：automatic summarization system; discussion
analysis; information retrieval
