任务型对话系统是当前自然语言处理领域的研究热点,对话状态跟踪作为任务型对话系统的核心模块,其主要任务是维护对话的上下文信息并以特定的状态形式展现。目前基于多领域的任务型对话系统由于对话场景复杂,导致对话状态难以跟踪,预测精度不高。该文提出一种融合槽位关联和语义关联的状态跟踪模型DST-S2C(Dialogue State Tracking with Slot Connection and Semantic Connection)。该模型将槽位构建成多关系图,并利用层级图注意力网络对槽位关系进行建模,提取融合多种槽位关联信息的槽位向量。同时,在槽门机制中加入词级语义相似度向量作为增强特征,获得对话上下文与槽位的局部语义信息,提高槽门机制的预测精度。实验表明,相较于基线模型,DST-S2C在MultiWOZ 2.1数据集上,联合准确率和槽位准确率分别提升了1.12%和0.39%。
Abstract
Dialogue state tracking is the core module of task-oriented dialogue system. Currently, multi-domain task-oriented dialogue system is difficult to track the dialogue state due to the complex dialogue scene. This paper proposes a state tracking model named DST-S2C,i.e.dialogue state tracking with slot connection and semantic connection. The model constructs the slots into a multi-relational graph, uses the hierarchical graph attention network to model the slot relationship, and extracts the slot embeddings that fuses multiple related-slot information. Furthermore, the slot-gate mechanism adds the local semantic information between the dialogue context and slots, which is essential to enhance the slot-gate mechanism performance. Experiments on MultiWOZ2.1 datasets show that DST-S2C outperforms the baseline model by 1.12% in joint accuracy and 0.39% in slot accuracy.
关键词
任务型对话系统 /
对话状态跟踪 /
多领域
{{custom_keyword}} /
Key words
task-oriented dialogue system /
dialogue state tracking /
multi-domain
{{custom_keyword}} /
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] ZHANG Z, TAKANOBU R, ZHU Q, et al. Recent advances and challenges in task-oriented dialog systems[J]. Science China Technological Sciences, 2020, 63(10): 2011-2027.
[2] DAI Y, YU H, JIANG Y, et al. A survey on dialog management: Recent advances and challenges[J]. arXiv preprint arXiv: 2005.02233, 2020.
[3] BUDZIANOWSKI P, WEN T H, TSENG B H, et al. MultiWOZ: A large-scale multi-domain wizard-of-oz dataset for task-oriented dialogue modelling[C]//Proceedings of the Conference on Empirical Methods in Natural Language Processing, 2018: 5016-5026.
[4] ERIC M, GOEL R, PAUL S, et al. MultiWOZ 2.1: A consolidated multi-domain dialogue dataset with state corrections and state tracking baselines[C]//Proceedings of the 12th Language Resources and Evaluation Conference, 2020: 422-428.
[5] MRKSIC N, SEAGHDHA D , WEN T H, et al. Neural belief tracker: Data-driven dialogue state tracking[C]//Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics, 2017: 1777-1788.
[6] ZHONG V, XIONG C, SOCHER R. Global-locally self-attentive dialogue state tracker[C]//Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics, 2018: 1458-1467.
[7] DEVLIN J, CHANG M W, LEE K, et al. Bert: Pre-training of deep bidirectional transformers for language understanding[J]. arXiv preprint arXiv: 1810.04805, 2018.
[8] LEE H, LEE J, KIM T Y. SUMBT: Slot-utterance matching for universal and scalable belief tracking[C]//Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics,2019: 5478-5483.
[9] WU C S, MADOTTO A, HOSSEINIASL E, et al. Transferable multidomain state generator for task oriented dialogue systems[C]//Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, 2019: 808-819.
[10] SEE A, LIU P J, MANNING CD. Get to the point: Summarization with pointer-generator networks[C]//Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics, 2017: 1073-1083.
[11] CHO K, VAN B, G C, et al. Learning phrase representations using RNN encoder-decoder for statistical machine translation[J]. arXiv preprint arXiv: 1406.1078, 2014.
[12] MCCANN B, KES N S, XIONG C, et al. The natural language decathlon: Multitask learning as question answering[J]. arXiv preprint arXiv: 1806.08730, 2018.
[13] REN L, NI J, MCAULEY J. Scalable and accurate dialogue state tracking via hierarchical sequence generation[C]//Proceedings of the Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, 2019: 1876-1885.
[14] KIM S, YANG S, KIM G, et al. Efficient dialogue state tracking by selectively overwriting memory[C]//Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, 2020: 567-582.
[15] VELICKOVIC P, CUCURULL G, CAS A, et al. Graph attention networks[J]. arXiv preprint arXiv: 1710.10903, 2017.
[16] VASWANI A, SHAZEER N, PARMER N, et al. Attention is all you need[C]//Proceedings of the 31st International Conterence on Neural Information Processing Systems, 2017: 6000-6010.
[17] SHEN T, WANG X. Multidomain dialogue state tracking with hierarchical task graph[C]//Proceedings of the International Joint Conference on Neural Networks. IEEE, 2020: 1-8.
[18] CHEN L, LV B, WANG C, et al. Schema-guided multi-domain dialogue state tracking with graph attention neural networks[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2020, 34(5): 7521-7528.
[19] ZHU S, LI J, CHEN L, et al. Efficient context and schema fusion networks for multi-domain dialogue state tracking[C]//Proceedings of the Association for Computational Linguistics, 2020: 766-781.
[20] WU P, ZOU B, JIANG R, et al. GCDST: A graph-based and copyaugmented multidomain dialogue state tracking[C]//Proceedings of the Association for Computational Linguistics, 2020.
[21] KIPF T N, WELLING M. Semi-supervised classification with graph convolutional networks[J]. arXiv preprint arXiv: 1609.02907, 2016.
{{custom_fnGroup.title_cn}}
脚注
{{custom_fn.content}}
基金
国防基础科研基金(JCKY2020605C003)
{{custom_fund}}
PDF(1815 KB)
