Article Structure
Abstract
Automated summarization methods can be defined as "language-independent,” if they are not based on any language-specific knowledge.
Introduction
Document summaries should use a minimum number of words to express a document’s main ideas.
Related Work
Extractive summarization is aimed at the selection of a subset of the most relevant fragments from a source text into the summary.
MUSE — MUltilingual Sentence Extractor
In this paper we propose a learning approach to language-independent extractive summarization where the best set of weights for a linear combination of sentence scoring methods is found by a genetic algorithm trained on a collection of document summaries.
Experiments
4.1 Overview
Conclusions and future work
In this paper we introduced MUSE, a new, GA-based approach to multilingual extractive summarization.
Topics
graph-based
- Today, graph-based text representations are becoming increasingly popular, due to their ability to enrich the document model with syntactic and semantic relations.Page 3, “Related Work”
- (1997) were among the first to make an attempt at using graph-based ranking methods in single document extractive summarization, generating similarity links between document paragraphs and using degree scores in order to extract the important paragraphs from the text.Page 3, “Related Work”
- Erkan and Radev (2004) and Mihalcea (2005) introduced algorithms for unsupervised extractive summarization that rely on the application of iterative graph-based ranking algorithms, such as PageRank (Erin and Page, 1998) and HITS (Kleinberg, 1999).Page 3, “Related Work”
- In contrast, representation used by the graph-based methods (except for TextRank) is based on the word-based graph representation models described in (Schenker et al., 2004).Page 3, “MUSE — MUltilingual Sentence Extractor”
See all papers in Proc. ACL 2010 that mention graph-based.
See all papers in Proc. ACL that mention graph-based.
Back to top.
cross validation
- We estimated the ROUGE metric using 10-fold cross validation .Page 7, “Experiments”
- Each corpus was then subjected to 10-fold cross validation , and the average results for training and testing were calculated.Page 7, “Experiments”
- Table 3: Results of 10-fold cross validation ENG HEB MULT Train 0.4483 0.5993 0.5205 Test 0.4461 0.5936 0.5027Page 8, “Experiments”
See all papers in Proc. ACL 2010 that mention cross validation.
See all papers in Proc. ACL that mention cross validation.
Back to top.
