Lune

EMNLP Findings2021

Keyphrase Generation with Fine-Grained Evaluation-Guided Reinforcement Learning

Yichao Luo, Yige Xu, Jiacheng Ye, Xipeng Qiu, Qi Zhang

Abstract

Aiming to generate a set of keyphrases, Keyphrase Generation (KG) is a classical task for capturing the central idea from a given document. Based on Seq2Seq models, the previous reinforcement learning framework on KG tasks utilizes the evaluation metrics to further improve the well-trained neural models. However, these KG evaluation metrics such as F 1 @5 and F 1 @M are only aware of the exact correctness of predictions on phrase-level and ignore the semantic similarities between similar predictions and targets, which inhibits the model from learning deep linguistic patterns. In response to this problem, we propose a new fine-grained evaluation metric to improve the RL framework, which considers different granularities: token-level F 1 score, edit distance, duplication, and prediction quantities. On the whole, the new framework includes two reward functions: the fine-grained evaluation score and the vanilla F 1 score. This framework helps the model identifying some partial match phrases which can be further optimized as the exact match ones. Experiments on KG benchmarks show that our proposed training framework outperforms the previous RL training frameworks among all evaluation scores. In addition, our method can effectively ease the synonym problem and generate a higher quality prediction. The source code is available at https://github.com/xuyige/FGRL4KG.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

2 top-tier papers have cited it since. Know what they found before your reviewers do.

Questions to start from

Lune can draw on

the full paper · the top-tier papers that cite it · related work

Get the answer

Free to start. No credit card required.

Lineage

  • One2Set: Generating Diverse Keyphrases as a Set cites One Size Does Not Fit All: Generating and Evaluating Variable Number of Keyphrases
  • One2Set: Generating Diverse Keyphrases as a Set cites Exclusive Hierarchical Decoding for Deep Keyphrase Generation
  • Rethinking Model Selection and Decoding for Keyphrase Generation with Pre-trained Sequence-to-Sequence Models cites One2Set: Generating Diverse Keyphrases as a Set
  • Rethinking Model Selection and Decoding for Keyphrase Generation with Pre-trained Sequence-to-Sequence Models cites One Size Does Not Fit All: Generating and Evaluating Variable Number of Keyphrases
  • Rethinking Model Selection and Decoding for Keyphrase Generation with Pre-trained Sequence-to-Sequence Models cites Exclusive Hierarchical Decoding for Deep Keyphrase Generation
  • Heterogeneous Graph Neural Networks for Keyphrase Generation cites One2Set: Generating Diverse Keyphrases as a Set
  • Heterogeneous Graph Neural Networks for Keyphrase Generation cites One Size Does Not Fit All: Generating and Evaluating Variable Number of Keyphrases
  • Heterogeneous Graph Neural Networks for Keyphrase Generation cites Exclusive Hierarchical Decoding for Deep Keyphrase Generation
  • Exclusive Hierarchical Decoding for Deep Keyphrase Generation cites One Size Does Not Fit All: Generating and Evaluating Variable Number of Keyphrases
Builds onCited byLeft to right by year; the more cited, the higher.

Cited by top-tier papers2

  1. Heterogeneous Graph Neural Networks for Keyphrase GenerationJiacheng Ye, Ruijian Cai, Tao Gui, Qi ZhangEMNLP 2021 · 14 citations
  2. Rethinking Model Selection and Decoding for Keyphrase Generation with Pre-trained Sequence-to-Sequence ModelsDi Wu, Wasi Uddin Ahmad, Kai-Wei ChangEMNLP 2023 · 6 citations

Builds on3

  1. One Size Does Not Fit All: Generating and Evaluating Variable Number of KeyphrasesXingdi Yuan, Tong Wang, Rui Meng, Khushboo Thaker et al.ACL 2020 · 76 citations
  2. Exclusive Hierarchical Decoding for Deep Keyphrase GenerationWang Chen, Hou Pong Chan, Piji Li, Irwin KingACL 2020 · 62 citations
  3. One2Set: Generating Diverse Keyphrases as a SetJiacheng Ye, Tao Gui, Yichao Luo, Yige Xu et al.ACL 2021

Related papers

  1. SimCKP: Simple Contrastive Learning of Keyphrase RepresentationsMinseok Choi, Chaeheon Gwak, Seho Kim, Si Hyeong Kim et al.EMNLP Findings 2023 · 3 citations
  2. HTKG: Deep Keyphrase Generation with Neural Hierarchical Topic GuidanceYuxiang Zhang, Tao Jiang, Tianyu Yang, Xiaoli Li et al.SIGIR 2022 · 14 citations
  3. Adaptive Beam Search Decoding for Discrete Keyphrase GenerationXiaoli Huang, Tongge Xu, Lvan Jiao, Yueran Zu et al.AAAI 2021 · 10 citations
  4. KTAE: A Model-Free Algorithm to Key-Tokens Advantage Estimation in Mathematical ReasoningWei Sun, Wen Yang, Pu Jian, Qianlong Du et al.NeurIPS 2025 · 22 citations
  5. KPEval: Towards Fine-Grained Semantic-Based Keyphrase EvaluationDi Wu, Da Yin, Kai-Wei ChangACL Findings 2024 · 7 citations
Dusk over the sea between two cliffs drawn in fine vertical lines