Krimping texts for better summarization

Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, pages 1931–1935, Lisbon, Portugal, 17-21 September 2015. c⃝2015 Association for Computational Linguistics. Krimping texts for better summarization Marina Litvak Sami Shamoon College of Engineering, Beer Sheva, Israel marinal@sce.ac.il Natalia Vanetik Sami Shamoon College of Engineering, Beer Sheva, Israel natalyav@sce.ac.il Mark Last Ben-Gurion University of the Negev, Beer Sheva, Israel mlast@bgu.ac.il Abstract Automated text summarization is aimed at extracting essential information from orig- inal text and presenting it in a minimal, often predefined, number of words. In this paper, we introduce a new approach for unsupervised extractive summariza- tion, based on the Minimum Description Length (MDL) principle, using the Krimp dataset compression algorithm (Vreeken et al., 2011). Our approach represents a text as a transactional dataset, with sen- tences as transactions, and then describes it by itemsets that stand for frequent se- quences of words. The summary is then compiled from sentences that compress (and as such, best describe) the document. The problem of summarization is reduced to the maximal coverage, following the assumption that a summary that best de- scribes the original text, should cover most of the word sequences describing the doc- ument. We solve it by a greedy algorithm and present the evaluation results. 1 Introduction Many unsupervised approaches for extractive text summarization follow the maximal coverage prin- ciple (Takamura and Okumura, 2009; Gillick and Favre, 2009), where the extract that maximally covers the information contained in the source text, is selected. Since the exhaustive solution de- mands an exponential number of tests, approxima- tion techniques, such as a greedy approach or a global optimization of a target function, are uti- lized. It is quite common to measure text infor- mativeness by the frequency of its components– words, phrases, concepts, and so on. A different approach that received much less atten- tion is based on the Minimum Description Length (MDL) principle, defining the best summary as the one that leads to the best compression of the text by providing its shortest and most concise descrip- tion. The MDL principle is widely useful in com- pression techniques of non-textual data, such as summarization of query results for OLAP appli- cations. (Lakshmanan et al., 2002; Bu et al., 2005) However, only a few works on text summarization using MDL can be found in the literature. Authors of (Nomoto and Matsumoto, 2001) used K-means clustering extended with MDL principle for find- ing diverse topics in the summarized text. Nomoto in (2004) also extended the C4.5 classifier with MDL for learning rhetorical relations. In (Nguyen et al., 2015) the problem of micro-review summa- rization is formulated within the MDL framework, where the authors view the tips as being encoded by snippets, and seek to find a collection of snip- pets that produce the encoding with the minimum number of bits. This paper introduces a new MDL-based approach for extracting relevant sentences into a summary. The approach represents documents as a sequen- tial transactional dataset and then compresses it by replacing …

Content truncated to 500 words