基于语句重写的无监督对话主题分割模型

Xia Hou, Qifeng Li, Tongliang Li
{"title":"基于语句重写的无监督对话主题分割模型","authors":"Xia Hou, Qifeng Li, Tongliang Li","doi":"arxiv-2409.07672","DOIUrl":null,"url":null,"abstract":"Dialogue topic segmentation plays a crucial role in various types of dialogue\nmodeling tasks. The state-of-the-art unsupervised DTS methods learn topic-aware\ndiscourse representations from conversation data through adjacent discourse\nmatching and pseudo segmentation to further mine useful clues in unlabeled\nconversational relations. However, in multi-round dialogs, discourses often\nhave co-references or omissions, leading to the fact that direct use of these\ndiscourses for representation learning may negatively affect the semantic\nsimilarity computation in the neighboring discourse matching task. In order to\nfully utilize the useful cues in conversational relations, this study proposes\na novel unsupervised dialog topic segmentation method that combines the\nUtterance Rewriting (UR) technique with an unsupervised learning algorithm to\nefficiently utilize the useful cues in unlabeled dialogs by rewriting the\ndialogs in order to recover the co-referents and omitted words. Compared with\nexisting unsupervised models, the proposed Discourse Rewriting Topic\nSegmentation Model (UR-DTS) significantly improves the accuracy of topic\nsegmentation. The main finding is that the performance on DialSeg711 improves\nby about 6% in terms of absolute error score and WD, achieving 11.42% in terms\nof absolute error score and 12.97% in terms of WD. on Doc2Dial the absolute\nerror score and WD improves by about 3% and 2%, respectively, resulting in SOTA\nreaching 35.17% in terms of absolute error score and 38.49% in terms of WD.\nThis shows that the model is very effective in capturing the nuances of\nconversational topics, as well as the usefulness and challenges of utilizing\nunlabeled conversations.","PeriodicalId":501030,"journal":{"name":"arXiv - CS - Computation and Language","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2024-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"An Unsupervised Dialogue Topic Segmentation Model Based on Utterance Rewriting\",\"authors\":\"Xia Hou, Qifeng Li, Tongliang Li\",\"doi\":\"arxiv-2409.07672\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Dialogue topic segmentation plays a crucial role in various types of dialogue\\nmodeling tasks. The state-of-the-art unsupervised DTS methods learn topic-aware\\ndiscourse representations from conversation data through adjacent discourse\\nmatching and pseudo segmentation to further mine useful clues in unlabeled\\nconversational relations. However, in multi-round dialogs, discourses often\\nhave co-references or omissions, leading to the fact that direct use of these\\ndiscourses for representation learning may negatively affect the semantic\\nsimilarity computation in the neighboring discourse matching task. In order to\\nfully utilize the useful cues in conversational relations, this study proposes\\na novel unsupervised dialog topic segmentation method that combines the\\nUtterance Rewriting (UR) technique with an unsupervised learning algorithm to\\nefficiently utilize the useful cues in unlabeled dialogs by rewriting the\\ndialogs in order to recover the co-referents and omitted words. Compared with\\nexisting unsupervised models, the proposed Discourse Rewriting Topic\\nSegmentation Model (UR-DTS) significantly improves the accuracy of topic\\nsegmentation. The main finding is that the performance on DialSeg711 improves\\nby about 6% in terms of absolute error score and WD, achieving 11.42% in terms\\nof absolute error score and 12.97% in terms of WD. on Doc2Dial the absolute\\nerror score and WD improves by about 3% and 2%, respectively, resulting in SOTA\\nreaching 35.17% in terms of absolute error score and 38.49% in terms of WD.\\nThis shows that the model is very effective in capturing the nuances of\\nconversational topics, as well as the usefulness and challenges of utilizing\\nunlabeled conversations.\",\"PeriodicalId\":501030,\"journal\":{\"name\":\"arXiv - CS - Computation and Language\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-09-12\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - CS - Computation and Language\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2409.07672\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - CS - Computation and Language","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2409.07672","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0

摘要

对话主题分割在各类对话建模任务中发挥着至关重要的作用。最先进的无监督 DTS 方法通过相邻话语匹配和伪分段,从对话数据中学习话题分段话语表征,从而进一步挖掘未标记对话关系中的有用线索。然而,在多轮对话中,话语经常会有共指或遗漏,导致直接使用这些话语进行表征学习可能会对相邻话语匹配任务中的语义相似性计算产生负面影响。为了充分利用对话关系中的有用线索,本研究提出了一种新型的无监督对话主题分割方法,该方法结合了语篇重写(UR)技术和无监督学习算法,通过重写对话来恢复共指词和遗漏词,从而有效地利用未标记对话中的有用线索。与现有的无监督模型相比,所提出的话语重写主题分割模型(UR-DTS)显著提高了主题分割的准确性。主要发现是,在 DialSeg711 上,绝对错误分值和 WD 的性能提高了约 6%,绝对错误分值提高了 11.42%,WD 提高了 12.97%;在 Doc2Dial 上,绝对错误分值和 WD 分别提高了约 3% 和 2%,SOTA 达到 35.这表明该模型在捕捉对话主题的细微差别以及利用无标记对话的实用性和挑战方面非常有效。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
查看原文
分享 分享
微信好友 朋友圈 QQ好友 复制链接
本刊更多论文
An Unsupervised Dialogue Topic Segmentation Model Based on Utterance Rewriting
Dialogue topic segmentation plays a crucial role in various types of dialogue modeling tasks. The state-of-the-art unsupervised DTS methods learn topic-aware discourse representations from conversation data through adjacent discourse matching and pseudo segmentation to further mine useful clues in unlabeled conversational relations. However, in multi-round dialogs, discourses often have co-references or omissions, leading to the fact that direct use of these discourses for representation learning may negatively affect the semantic similarity computation in the neighboring discourse matching task. In order to fully utilize the useful cues in conversational relations, this study proposes a novel unsupervised dialog topic segmentation method that combines the Utterance Rewriting (UR) technique with an unsupervised learning algorithm to efficiently utilize the useful cues in unlabeled dialogs by rewriting the dialogs in order to recover the co-referents and omitted words. Compared with existing unsupervised models, the proposed Discourse Rewriting Topic Segmentation Model (UR-DTS) significantly improves the accuracy of topic segmentation. The main finding is that the performance on DialSeg711 improves by about 6% in terms of absolute error score and WD, achieving 11.42% in terms of absolute error score and 12.97% in terms of WD. on Doc2Dial the absolute error score and WD improves by about 3% and 2%, respectively, resulting in SOTA reaching 35.17% in terms of absolute error score and 38.49% in terms of WD. This shows that the model is very effective in capturing the nuances of conversational topics, as well as the usefulness and challenges of utilizing unlabeled conversations.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
自引率
0.00%
发文量
0
期刊最新文献
LLMs + Persona-Plug = Personalized LLMs MEOW: MEMOry Supervised LLM Unlearning Via Inverted Facts Extract-and-Abstract: Unifying Extractive and Abstractive Summarization within Single Encoder-Decoder Framework Development and bilingual evaluation of Japanese medical large language model within reasonably low computational resources Human-like Affective Cognition in Foundation Models
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
现在去查看 取消
×
提示
确定
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
已复制链接
已复制链接
快去分享给好友吧!
我知道了
×
扫码分享
扫码分享
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 联系我们:info@booksci.cn
Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。
Copyright © 2023 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1