Aspect-Driven News Summarization
A summary of any event type is only complete if certain in-formation aspects are mentioned. For a court trial, readers will at leastwant to know who is involved and what the charges and the sentence are.For a natural disaster, they will ask for the disaster type, the victims andother damages. Will a co-occurrence or frequency-based sentence extrac-tion summariser automatically provide the requested information, or arethe results better if an information extraction (IE) system rst detectsthe summary-crucial aspects? To answer this question, we compared theperformance of a purely co-occurrence-based method with a system thatadditionally makes use of targeted IE. As each event type requires di er-ent information aspects and not all of them were covered by the existingIE software, we used a tool that learns semantically related terms to coverthe remaining aspects. The comprehensive evaluation in the TAC'2010competition showed that event extraction is indeed bene cial for sum-marisation performance, and that summary quality is directly related toIE quality. Our integrated system was ranked among the top systemsparticipating at TAC.
Year: 2011
Full text [225 kB]Authors of this publication:
Josef Steinberger
E-mail: jstein@kiv.zcu.cz
Related Projects:
Automatic Text Summarisation | |
| Authors: | Josef Steinberger, Karel Ježek, Michal Campr, Jiří Hynek |
| Desc.: | Automatic text summarisation using various text mining methods, mainly Latent Semantic Analysis (LSA). |