Paper: Learning Summary Content Units with Topic Modeling

ACL ID C10-2045
Title Learning Summary Content Units with Topic Modeling
Venue International Conference on Computational Linguistics
Session Poster Session
Year 2010
Authors

In the field of multi-document summa- rization, the Pyramid method has be- come an important approach for evaluat- ing machine-generated summaries. The method is based on the manual annotation of text spans with the same meaning in a set of human model summaries. In this paper, we present an unsupervised, prob- abilistic topic modeling approach for au- tomatically identifying such semantically similar text spans. Our approach reveals some of the structure of model summaries and identifies topics that are good approx- imations of the Summary Content Units (SCU) used in the Pyramid method. Our results show that the topic model identi- fies topic-sentence associations that corre- spond to the contributors of SCUs, sug- gesting that the topic modeling approach can generate a viable set of cand...