SummaRuNNer: A Recurrent Neural Network based Sequence Model for Extractive Summarization of Documents
This work addresses the problem of generating summaries from documents for users needing quick insights, offering an interpretable and label-efficient approach.
The authors tackled extractive document summarization by introducing SummaRuNNer, a recurrent neural network model that achieved performance comparable to state-of-the-art methods, with the added benefit of interpretability through feature visualization.
We present SummaRuNNer, a Recurrent Neural Network (RNN) based sequence model for extractive summarization of documents and show that it achieves performance better than or comparable to state-of-the-art. Our model has the additional advantage of being very interpretable, since it allows visualization of its predictions broken up by abstract features such as information content, salience and novelty. Another novel contribution of our work is abstractive training of our extractive model that can train on human generated reference summaries alone, eliminating the need for sentence-level extractive labels.