Development of automatic trend exploration system using MuST data collection

被引:0
作者
Murata, Masaki [1 ]
Ichii, Koji [2 ]
Ma, Qing [1 ,3 ]
Shirado, Tamotsu [1 ]
Kanamaru, Toshiyuki [4 ]
Tsukawaki, Sachiyo [1 ]
Isahara, Hitoshi [1 ]
机构
[1] Natl Inst Informat & Commun Technol, Knowledge Creating Commun Res Ctr, Seika, Kyoto 6190289, Japan
[2] Hiroshima Univ, Grad Sch Engn, Hiroshima 7398527, Japan
[3] Ryukoku Univ, Fac Sci & Technol, Shiga 5202194, Japan
[4] Kyoto Univ, Grad Sch Human & Environm Studies, Sakyo Ku, Kyoto 6068501, Japan
来源
INFORMATION-AN INTERNATIONAL INTERDISCIPLINARY JOURNAL | 2008年 / 11卷 / 06期
关键词
trend information exploration system; unit expression; graph; highlighting; sentence extraction;
D O I
暂无
中图分类号
T [工业技术];
学科分类号
08 ;
摘要
We studied the automatic extraction of trend information from texts, such as newspaper articles. This kind of study is useful for exploring and examining trends. We used data sets provided by a workshop on multimodal summarization for trend information (MuST Workshop) to construct our automatic trend exploration system. In this system, we first extract units, temporals, and item expressions from newspaper articles. Next, we extract pairs of expressions as trend information. Finally, we arrange the pairs and display them in graphs. When we judged an extraction of a correct graph in the top output in the experiments to be correct, our system obtained 0.25 in evaluation A and 0.33 in evaluation B. When we judged the extraction of a correct graph in the top five outputs to be correct, it obtained 0.42 in evaluation A and 0.63 in evaluation B. Evaluation A is defined as a graph where 75% or more of the dots are correct is judged to be correct. Evaluation B is defined as a graph where 50% or more of the dots are correct is judged to be correct. Our system is convenient and effective, because it can output a graph that includes trend information at these types of accuracy rates by only giving a set of documents as an input.
引用
收藏
页码:811 / 827
页数:17
相关论文
共 4 条
[1]  
Kato T., 2005, P 5 NTCIR WORKSH M E
[2]  
MATSUMOTO Y, 1999, JAPANESE MORPHOLOGIC
[3]  
MURATA M, 2005, J NATURAL LANGUAGE, V12, P209
[4]  
ROBERTSON S, 1994, TREC 3