Deep Communicating Agents for Abstractive Summarization
This addresses the problem of generating coherent summaries from long texts for natural language processing applications, with incremental improvements over existing methods.
The paper tackles the challenge of representing long documents for abstractive summarization by using deep communicating agents in an encoder-decoder architecture, resulting in higher quality summaries compared to strong baselines.
We present deep communicating agents in an encoder-decoder architecture to address the challenges of representing a long document for abstractive summarization. With deep communicating agents, the task of encoding a long text is divided across multiple collaborating agents, each in charge of a subsection of the input text. These encoders are connected to a single decoder, trained end-to-end using reinforcement learning to generate a focused and coherent summary. Empirical results demonstrate that multiple communicating encoders lead to a higher quality summary compared to several strong baselines, including those based on a single encoder or multiple non-communicating encoders.