Interactive Document Summarization

Raoufdine Said, Adrien Guille

Abstract: With the advent of modern chatbots, automatic summarization is becoming common practice to quicken access to information. However the summaries they generate can be biased, unhelpful or untruthful. Hence, in sensitive scenarios, extractive summarization remains a more reliable approach. In this paper we present an original extractive method combining a GNN-based encoder and a RNN-based decoder, coupled with a user-friendly interface that allows for interactive summarization.