Cross-language document summarization via extraction and ranking of multiple summaries

Abstract

The task of cross-language document summarization aims to produce a summary in a target language (e.g., Chinese) for a given document set in a different source language (e.g., English). Previous studies focus on ranking and selection of translated sentences in the target language. In this paper, we propose a new framework for addressing the task by extraction and ranking of multiple summaries in the target language. First, we extract multiple candidate summaries by proposing several schemes for improving the upper-bound quality of the summaries. Then, we propose a new ensemble ranking method for ranking the candidate summaries by making use of bilingual features. Extensive experiments have been conducted on a benchmark dataset and the results verify the effectiveness of our proposed framework, which outperforms a variety of baselines, including supervised baselines.