Cross-Language Text Summarization Using Sentence and Multi-Sentence Compression

Abstract

Abstract. Cross-Language Automatic Text Summarization produces a summary in a language different from the language of the source documents. In this paper, we propose a French-to-English cross-lingual summarization framework that analyzes the information in both languages to identify the most relevant sentences. In order to generate more informative cross-lingual summaries, we introduce the use of chunks and two compression methods at the sentence and multi-sentence levels. Experimental results on the MultiLing 2011 dataset show that our framework improves the results obtained by state-of-the art approaches according to ROUGE metrics.

Cross-Lingual Speech-to-Text Summarization

“…The relevance of concepts and facts are estimated using the CoRank algorithm [25], while summaries were produced by fusing the most relevant source-side PAS elements considering their translation quality. Finally, a French-to-English cross-lingual abstractive summarization approach was proposed in [12]. This CLTS system combined multi-sentence and sentence compression methods in order to produce informative cross-lingual summaries.…”

“…French-to-English CLTS aims to generate an English short summary that describes the main information from a French transcript document. Following the CLTS approach proposed by Linhares Pontes et al [12], we analyze documents in the source and the target languages to select the most relevant sentences. Then, we create clusters of similar sentences to independently analyze the subjects of a document.…”

“…Following the idea proposed in [12], we consider the similarity in both languages to create clusters of similar sentences. Then, we use the Stanford CoreNLP tool [15] with jMWE [9] to detect Multi-Word Expressions in the English side, while the corresponding expressions were deduced on the French side with the help of the Giza++ alignment tool [19].…”

“…Then, we use the Stanford CoreNLP tool [15] with jMWE [9] to detect Multi-Word Expressions in the English side, while the corresponding expressions were deduced on the French side with the help of the Giza++ alignment tool [19]. Among several state-of-the-art Multi-Sentence Compression (MSC) methods [4,1,18,13], we use Linhares Pontes et al's approach [12] to generate a compressed sentence from each cluster of similar sentences. This system builds compressions controlled by the presence of keywords, to increase informativeness, and 3-grams, to ensure grammaticality.…”

Microblog Contextualization: Advantages and Limitations of a Multi-sentence Compression Approach

“…Several studies have analyzed Text Summarization (TS) and Cross-Language Text Summarization (CLTS) [10,11,7]. TS aims to create a short, accurate, and fluid summary of a longer text document; CLTS also generates a summary but the language of the summary is different from the language of the source documents.…”