Towards multi-lingual adaptability of subjective logic based document summarization: A case study with Chinese documents

Abstract

Due to the rapid advancement of the internet technology, there is proliferation of textual data, as a result of which automatic summarization has become one of the useful means of coping with the problem of information overload. These textual data are not just in English; thus researchers started focusing on multilingual summarization platforms, so that single framework can be used to cope with different languages. Chinese being another widely spoken language, in this paper, we present an extension of Subjective Logic summarization framework (SubSum) [1], for Chinese. SubSum extracts significant sentences from documents to form extractive summaries. Quantifying uncertainty is the key advantage of SubSum over commonly used summarization methods. The main aim of this work is to show how well SubSum can be adapted to a completely different language, without making changes to the core framework. Moreover, extensive experiments on benchmark datasets demonstrate the effectiveness of SubSum applied for Chinese summarization.