Dialogue Summarization Combining Prompt Signal and Graph Structure

doi:10.3778/j.issn.1002-8331.2405-0101

Abstract

Abstract: With the increasing popularity of dialog-based communication, more and more researchers pay attention to the task of dialogue summarization, which aims to compress the complex dialogue text into a concise representation. In a dialogue text, the communication between multiple roles usually involves key information about a particular event and this information is distributed in several conversations. However, the existing methods cannot deeply study the internal relationship and structure of dialogue text, and it is easy to miss key information. To solve these problems, a dialogue summarization model combining prompt signal and graph structure is designed to help understand the dialogue structure and grasp the key information, so as to improve the accuracy of the summary. Based on prompt learning, discrete prompt signals are introduced into the prompt encoder to help the model focus on key information (keywords, topics, etc. ) more pertinently. At the same time, the dynamic graph structure is introduced to capture and integrate cross-sentence information by using the structural information of dialogue. Experimental results on SAMSum, QMsum and DialogSum datasets show that the ROUGE-1, ROUGE-2 and ROUGE-L scores are significantly improved, which verifies the effectiveness of the model proposed in this paper.

Key words: dialogue summarization, prompt learning, prompt signal, graph structure

摘要： 以对话形式为主的通信方式逐渐普及，对话摘要任务引起越来越多研究者的关注，该任务旨在将复杂的对话文本压缩成简洁的表示形式。在对话文本中，多个对话者之间的交流通常涉及有关某个特定事件的关键信息，且这些信息分布较为分散。然而，现有方法未深入研究对话内容的内在关系和结构，容易遗漏关键信息。针对上述问题，设计了结合提示信号与图结构的对话摘要生成模型，旨在帮助理解对话结构并把握对话中的关键信息，进而提高摘要生成的准确率。该模型基于提示学习引入了离散提示信号，并将其输入提示编码器，旨在利用提示信号协助模型更有针对性地聚焦对话的关键信息（关键词、主题词等）。同时，该模型引入动态图结构，旨在利用对话的结构性信息来捕捉并整合跨句子信息。在SAMSum、QMsum和DialogSum数据集上的实验结果表明，ROUGE-1、ROUGE-2和ROUGE-L得分均取得了显著提升，验证了模型的有效性。

关键词: 对话摘要, 提示学习, 提示信号, 图结构

JIN Yanliang, FENG Qiuyan, GAO Yuan. Dialogue Summarization Combining Prompt Signal and Graph Structure[J]. Computer Engineering and Applications, 2025, 61(15): 241-250.

金彦亮, 冯湫燕, 高塬. 结合提示信号与图结构的对话摘要生成模型[J]. 计算机工程与应用, 2025, 61(15): 241-250.

References

[1] 刘迪, 奚雪峰, 崔志明, 等. 抽取-生成式自动文本摘要技术研究综述[J]. 计算机技术与发展, 2023, 33(5): 1-8.
LIU D, XI X F, CUI Z M, et al. Review of research on extractive-abstractive automatic text summarization technology[J]. Computer Technology and Development, 2023, 33(5): 1-8.
[2] MOHAN M J, SUNITHA C, GANESH A, et al. A study on ontology based abstractive summarization[J]. Procedia Computer Science, 2016, 87: 32-37.
[3] SAKHARE D Y, KUMAR R, JANMEDA S. Development of embedded platform for Sanskrit grammar-based document summarization[M]//Speech and language processing for human-machine communications. Singapore: Springer, 2018: 41-50.
[4] ZHANG Y S, NI A S, YU T, et al. An exploratory study on long dialogue summarization: what works and what’s next[C]//Findings of the Association for Computational Linguistics. Stroudsburg: ACL, 2021: 4426-4433.
[5] 李健智, 王红玲, 王中卿. 基于场景与对话结构的摘要生成研究[J]. 计算机工程, 2023, 49(4): 303-311.
LI J Z, WANG H L, WANG Z Q. Research on summarization generation based on scene and dialogue structure[J]. Computer Engineering, 2023, 49(4): 303-311.
[6] GOO C W, CHEN Y N. Abstractive dialogue summarization with sentence-gated modeling optimized by dialogue acts[C]//Proceedings of the 2018 IEEE Spoken Language Technology Workshop. Piscataway: IEEE, 2018: 735-742.
[7] LIU C, WANG P, XU J, et al. Automatic dialogue summary generation for customer service[C]//Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, Anchorage, AK, USA, 2019: 1957-1965.
[8] NARAYAN S, ZHAO Y, MAYNEZ J, et al. Planning with learned entity prompts for abstractive summarization[J]. Transactions of the Association for Computational Linguistics, 2021, 9: 1475-1492.
[9] LIU Z Y, NG A, LEE S, et al. Topic-aware pointer-generator networks for summarizing spoken conversations[C]//Proceedings of the 2019 IEEE Automatic Speech Recognition and Understanding Workshop. Piscataway: IEEE, 2019: 814-821.
[10] SHIN J, YU H, MOON H, et al. Dialogue summaries as dialogue states (DS2), template-guided summarization for few-shot dialogue state tracking[C]//Findings of the Association for Computational Linguistics. Stroudsburg: ACL, 2022: 3824-3846.
[11] FENG Xiachong, FENG Xiaocheng, QIN B. Incorporating commonsense knowledge into abstractive dialogue summarization via heterogeneous graph networks[C]//Proceedings of the 20th Chinese National Conference on Computational Linguistics, Huhhot, China, 2021: 964-975.
[12] FERNANDES P, ALLAMANIS M, BROCKSCHMIDT M. Structured neural summarization[J]. arXiv:1811.01824, 2018.
[13] GLIWA B, MOCHOL I, BIESEK M, et al. SAMSum Corpus: a human-annotated dialogue dataset for abstractive summarization[J]. arXiv:1911.12237, 2019.
[14] ZHAO L, XU W, GUO J. Improving abstractive dialogue summarization with graph structures and topic words[C]//Proceedings of the 28th International Conference on Computational Linguistics. Spain: International Committee on Computational Linguistics, 2020: 437-449.
[15] CHEN J, YANG D. Multi-view sequence-to-sequence models with conversational structure for abstractive dialogue summarization[C]//Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing. Stroudsburg: Association for Computational Linguistics, 2020: 4106-4118.
[16] LECLAIR A, HAQUE S, WU L, et al. Improved code summarization via a graph neural network[C]//Proceedings of the 28th International Conference on Program Comprehension. New York: Association for Computing Machinery, 2020: 184-195.
[17] WANG D, LIU P, ZHENG Y, et al. Heterogeneous graph neural networks for extractive document summarization[C]//Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. Stroudsburg: Association for Computational Linguistics, 2020: 6209-6219.
[18] ZHAO L L, ZENG W H, XU W R, et al. Give the truth: incorporate semantic slot into abstractive dialogue summarization[C]//Findings of the Association for Computational Linguistics. Stroudsburg: ACL, 2021: 2435-2446.
[19] FENG Xiachong, FENG Xiaocheng, QIN B, et al. Dialogue discourse-aware graph model and data augmentation for meeting summarization[C]//Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence, 2021: 3808-3814.
[20] GAO S, CHENG X, LI M Z, et al. Dialogue summarization with static-dynamic structure fusion graph[C]//Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics. Stroudsburg: ACL, 2023: 13858-13873.
[21] 陈明轩, 肖诗斌, 王洪俊. 基于深度学习的生成式文本摘要综述[J]. 软件导刊, 2024, 23(5): 212-220.
CHEN M X, XIAO S B, WANG H J. A survey of deep learning-based generative text summarization[J]. Software Guide, 2024, 23(5): 212-220.
[22] LIU P F, YUAN W Z, FU J L, et al. Pre-train, prompt, and predict: a systematic survey of prompting methods in natural language processing[J]. ACM Computing Surveys, 2023, 55(9).
[23] DING N, HU S D, ZHAO W L, et al. OpenPrompt: an open-source framework for prompt-learning[C]//Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics: System Demonstrations. Stroudsburg: ACL, 2022: 105-113.
[24] DOU Z Y, LIU P F, HAYASHI H, et al. GSum: a general framework for guided neural abstractive summarization[C]//Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Stroudsburg: ACL, 2021: 4830-4842.
[25] LI L, ZHANG Y F, CHEN L. Personalized prompt learning for explainable recommendation[J]. ACM Transactions on Information Systems, 2023, 41(4): 1-26.
[26] DING N, CHEN Y, HAN X, et al. Prompt-learning for fine-grained entity typing[C]//Findings of the Association for Computational Linguistics. Stroudsburg: Association for Computational Linguistics, 2022: 6888-6901.
[27] NAIR V, SCHUMACHER E, KANNAN A. Generating medically-accurate summaries of patient-provider dialogue: a multi-stage approach using large language models[C]//Proceedings of the 5th Clinical Natural Language Processing Workshop. Stroudsburg: Association for Computational Linguistics, 2023: 200-217.
[28] XIE K, YU T, WANG H, et al. Few-shot dialogue summarization via skeleton-assisted prompt transfer[J]. arXiv:2305. 12077, 2023.
[29] ZHOU Y, RINGEVAL F, PORTET F. Can GPT models follow human summarization guidelines? Evaluating ChatGPT and GPT-4 for dialogue summarization[J]. arXiv:2310.16810, 2023.
[30] LEWIS M, LIU Y, GOYAL N, et al. BART: denoising sequence-to-sequence pre-training for natural language generation, translation, and comprehension[C]//Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. Stroudsburg: Association for Computational Linguistics, 2020: 7871-7880.
[31] MIHALCEA R, TARAU P. TextRank: brining order into texts[C]//Proceedings of the 2004 Conference on Empirical Methods in Natural Language Processing. Stroudsburg: Association for Computational Linguistics, 2004: 404-411.
[32] BLEI D M, NG A Y, JORDAN M I. Latent dirichlet allocation [J]. Journal of Machine Learning Research, 2003, 3(1): 993-1022.
[33] ZHONG M, YIN D, YU T, et al. QMSum: a new benchmark for query-based multi-domain meeting summarization[C]//Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Stroudsburg: ACL, 2021: 5905-5921.
[34] CHEN Y L, LIU Y, CHEN L, et al. DialogSum: a real-life scenario dialogue summarization dataset[C]//Findings of the Association for Computational Linguistics. Stroudsburg: ACL, 2021: 5062-5074.
[35] LIN C Y, HOVY E. Automatic evaluation of summaries using ngram co-occurrence statistics[C]//Proceedings of the 2003 Human Language Technology Conference of the North American Chapter of the Association for Computational Linguistics, 2003: 150-157.
[36] VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[C]//Advances in Neural Information Processing Systems, 2017.
[37] ZHU C, XU R, ZENG M, et al. A hierarchical network for abstractive meeting summarization with cross-domain pretraining[C]//Findings of the Association for Computational Linguistics. Stroudsburg: Association for Computational Linguistics, 2020: 194-203.
[38] CHEN J, YANG D. Structure-aware abstractive conversation summarization via discourse and action graphs[C]//Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Stroudsburg: Association for Computational Linguistics, 2021: 1380-1391.
[39] LIU Z, CHEN N F. Controllable neural dialogue summari-zation with personal named entity planning[C]//Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing. Stroudsburg: Association for Computational Linguistics, 2021: 92-106.
[40] YOO C, LEE H. Improving abstractive dialogue summarization using keyword extraction[J]. Applied Sciences, 2023, 13(17): 9771.
[41] LIU Z Y, WANG Z Y, WANG J H. A coarse-to-fine training paradigm forDialogue summarization[C]//Proceedings of International Conference on Artificial Neural Networks and Machine Learning. Cham: Springer, 2022: 416-427.
[42] ZHANG Z, LI J H. Topic-features for dialogue summarization[C]//Proceedings of the International Conference on Artificial Neural Networks Natural Language Processing and Chinese Computing. Cham: Springer International Publishing, 2022: 327-338.