PoBRL: Optimizing Multi-Document Summarization by Blending Reinforcement Learning Policies

2021/05/18 by Andy Su, Su, Andy, Difei Su +5
Computer Science · #Advanced Text Analysis Techniques #Artificial Intelligence (cs.AI) #FOS: Computer and information sciences #Natural Language Processing Techniques #Topic Modeling

paper · pdf · doi:10.48550/arxiv.2105.08244

openalex publication_date 2021/05/18 · openalex created_date 2021/05/24 · openalex updated_date 2026/07/28

Abstract

We propose a novel reinforcement learning based framework PoBRL for solving multi-document summarization. PoBRL jointly optimizes over the following three objectives necessary for a high-quality summary: importance, relevance, and length. Our strategy decouples this multi-objective optimization into different subproblems that can be solved individually by reinforcement learning. Utilizing PoBRL, we then blend each learned policies together to produce a summary that is a concise and complete representation of the original input. Our empirical analysis shows state-of-the-art performance on several multi-document datasets. Human evaluation also shows that our method produces high-quality output.

Citations

• A Deep Reinforced Model for Abstractive Summarization
• Graph-based Neural Multi-Document Summarization

Related

• Multi-Dimensional Optimization for Text Summarization via Reinforcement Learning
• Multi-document Summarization with Maximal Marginal Relevance-guided Reinforcement Learning
• Reward Learning for Efficient Reinforcement Learning in Extractive Document Summarisation
• Topic-Guided Reinforcement Learning with LLMs for Enhancing Multi-Document Summarization
• Balancing Rewards in Text Summarization: Multi-Objective Reinforcement Learning via HyperVolume Optimization