NCLS: Neural Cross-Lingual Summarization

TL;DR

This paper introduces NCLS, an end-to-end neural framework for cross-lingual summarization that outperforms pipeline methods and leverages multi-task learning with high-quality datasets generated via round-trip translation.

Contribution

It presents the first end-to-end neural cross-lingual summarization model and a novel multi-task training approach using large-scale datasets created through round-trip translation.

Findings

NCLS significantly outperforms traditional pipeline methods.

Multi-task learning further improves summary quality.

High-quality CLS datasets are effectively generated from monolingual data.

Abstract

Cross-lingual summarization (CLS) is the task to produce a summary in one particular language for a source document in a different language. Existing methods simply divide this task into two steps: summarization and translation, leading to the problem of error propagation. To handle that, we present an end-to-end CLS framework, which we refer to as Neural Cross-Lingual Summarization (NCLS), for the first time. Moreover, we propose to further improve NCLS by incorporating two related tasks, monolingual summarization and machine translation, into the training process of CLS under multi-task learning. Due to the lack of supervised CLS data, we propose a round-trip translation strategy to acquire two high-quality large-scale CLS datasets based on existing monolingual summarization datasets. Experimental results have shown that our NCLS achieves remarkable improvement over traditional pipeline methods on both English-to-Chinese and Chinese-to-English CLS human-corrected test sets. In addition, NCLS with multi-task learning can further significantly improve the quality of generated summaries. We make our dataset and code publicly available here: http://www.nlpr.ia.ac.cn/cip/dataset.htm.