Text-Mining Research Group

Knowledge-poor Multilingual Sentence Compression

We present a feature-based method for sentence compression. Firstly, a summary is created by our summarization method based on latent semantic analysis. The compression approach then removes unimportant clauses from the summary sentences. For each sentence a set of its possible compressed forms (compression candidates) is created. The candidates are then classified using 8 proposed features into two classes: in the first class there are candidates in which the important information was removed by compression and in the second class the information was still contained. The shortest candidate from the latter group substitutes the full sentence in the summary. The features are knowledge-poor which enables them to work with whatever language and the method can be easily extended by other features.

Keywords: Text summarization, sentence compression

Year: 2007

Download:  Full text [226 kB]

Josef Steinberger

E-mail: jstein@kiv.zcu.cz

Josef is an associated professor at the Department of computer science and engineering at the University of West Bohemia in Pilsen, Czech Republic. He is interested in media monitoring and analysis, mainly automatic text summarisation, sentiment analysis and coreference resolution.

Roman Tesař

Phone: +420 377632479
E-mail: roman.tesar@gmail.com
WWW: http://www.sweb.cz/romant1/CV.pdf

Roman is a PhD student at the Department of Computer Science and Engineering, Faculty of Applied Sciences, University of West Bohemia in Pilsen, Czech Republic. His work is focused on the utilization of word n-grams in text classification and document filtering.

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).