{"url":"/dataset/multiwoz","name":"MultiWOZ","full_name":"Multi-domain Wizard-of-Oz","description_markdown":"The **Multi-domain Wizard-of-Oz** (**MultiWOZ**) dataset is a large-scale human-human conversational corpus spanning over seven domains, containing 8438 multi-turn dialogues, with each dialogue averaging 14 turns. Different from existing standard datasets like WOZ and DSTC2, which contain less than 10 slots and only a few hundred values, MultiWOZ has 30 (domain, slot) pairs and over 4,500 possible values. The dialogues span seven domains: restaurant, hotel, attraction, taxi, train, hospital and police.\r\n\r\nSource: [Contents](https://arxiv.org/abs/1905.07687)\r\n\r\nImage Source: [Zhang et al](https://www.researchgate.net/figure/Example-of-the-difference-between-dialogue-state-annotation-in-MultiWOZ-21-and-MultiWOZ_fig2_343022084)","description_withheld":null,"homepage":"https://github.com/budzianowski/multiwoz","introduced_date":"2018-01-01","introduced_date_note":null,"introduced_by":{"paper":"/paper/multiwoz-a-large-scale-multi-domain-wizard-of","title":"MultiWOZ -- A Large-Scale Multi-Domain Wizard-of-Oz Dataset for Task-Oriented Dialogue Modelling","first_author":"Paweł Budzianowski","url":null},"license":{"name":"MIT","url":"https://github.com/budzianowski/multiwoz"},"modalities":[{"name":"Texts","url":"/datasets/modality/texts"}],"tasks":[{"name":"Data-to-Text Generation","url":"/task/data-to-text-generation","datasets_with_task":"/datasets/task/data-to-text-generation"},{"name":"Intent Detection","url":"/task/intent-detection","datasets_with_task":"/datasets/task/intent-detection"},{"name":"Slot Filling","url":"/task/slot-filling","datasets_with_task":"/datasets/task/slot-filling"},{"name":"Dialogue State Tracking","url":"/task/dialogue-state-tracking","datasets_with_task":"/datasets/task/dialogue-state-tracking"},{"name":"Task-Oriented Dialogue Systems","url":"/task/task-oriented-dialogue-systems","datasets_with_task":"/datasets/task/task-oriented-dialogue-systems"},{"name":"Multi-domain Dialogue State Tracking","url":"/task/multi-domain-dialogue-state-tracking","datasets_with_task":"/datasets/task/multi-domain-dialogue-state-tracking"},{"name":"End-To-End Dialogue Modelling","url":"/task/end-to-end-dialogue-modelling","datasets_with_task":"/datasets/task/end-to-end-dialogue-modelling"},{"name":"domain classification","url":"/task/domain-classification","datasets_with_task":"/datasets/task/domain-classification"}],"languages":[{"name":"English","url":"/datasets/language/english"}],"variants":["MULTIWOZ 2.0","MULTIWOZ 2.1","MultiWOZ","MULTIWOZ 2.2","MULTIWOZ 2.3","MULTIWOZ 2.4"],"data_loaders":[{"repo":"https://github.com/huggingface/datasets","url":"https://huggingface.co/datasets/pfb30/multi_woz_v22","frameworks":["tf","pytorch","jax"]},{"repo":"https://github.com/huggingface/datasets","url":"https://huggingface.co/datasets/multi_woz_v22","frameworks":["tf","pytorch","jax"]},{"repo":"https://github.com/facebookresearch/ParlAI","url":"https://parl.ai/docs/tasks.html#multiwoz-2.0","frameworks":["pytorch"]},{"repo":"https://github.com/budzianowski/multiwoz","url":"https://github.com/budzianowski/multiwoz","frameworks":["pytorch"]}],"num_papers_in_archive":328,"source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28"},"benchmarks":[{"leaderboard":"/sota/end-to-end-dialogue-modelling-on-multiwoz-2-0","task":"End-To-End Dialogue Modelling","dataset_variant":"MULTIWOZ 2.0","rows":6,"metrics":["MultiWOZ (Success)","MultiWOZ (Inform)","BLEU"],"first_row_in_archive_order":{"model":"GALAXY","paper":"/paper/galaxy-a-generative-pre-trained-model-for","metrics":{"BLEU":"20.5","MultiWOZ (Inform)":"94.4","MultiWOZ (Success)":"85.3"},"code_links":[{"title":"siat-nlp/galaxy","url":"https://github.com/siat-nlp/galaxy"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"},{"leaderboard":"/sota/data-to-text-generation-on-multiwoz-2-1","task":"Data-to-Text Generation","dataset_variant":"MULTIWOZ 2.1","rows":5,"metrics":["BLEU"],"first_row_in_archive_order":{"model":"T5-Base","paper":"/paper/text-to-text-pre-training-for-data-to-text","metrics":{"BLEU":"35.1"},"code_links":[{"title":"google-research-datasets/ToTTo","url":"https://github.com/google-research-datasets/ToTTo"},{"title":"shark-nlp/cont","url":"https://github.com/shark-nlp/cont"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"},{"leaderboard":"/sota/dialogue-state-tracking-on-multiwoz-2-1","task":"Dialogue State Tracking","dataset_variant":"MULTIWOZ 2.1","rows":4,"metrics":["Joint Acc","MultiWOZ (Joint Goal Acc)"],"first_row_in_archive_order":{"model":"DeepStruct multi-task w/ finetune","paper":"/paper/deepstruct-pretraining-of-language-models-for-1","metrics":{"Joint Acc":"54.2"},"code_links":[{"title":"cgraywang/deepstruct","url":"https://github.com/cgraywang/deepstruct"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"},{"leaderboard":"/sota/end-to-end-dialogue-modelling-on-multiwoz-2-1","task":"End-To-End Dialogue Modelling","dataset_variant":"MULTIWOZ 2.1","rows":4,"metrics":["BLEU","MultiWOZ (Inform)","MultiWOZ (Success)"],"first_row_in_archive_order":{"model":"GALAXY","paper":"/paper/galaxy-a-generative-pre-trained-model-for","metrics":{"BLEU":"20.01","MultiWOZ (Inform)":"95.30","MultiWOZ (Success)":"86.20"},"code_links":[{"title":"siat-nlp/galaxy","url":"https://github.com/siat-nlp/galaxy"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"},{"leaderboard":"/sota/dialogue-state-tracking-on-multiwoz-2-2","task":"Dialogue State Tracking","dataset_variant":"MULTIWOZ 2.2","rows":2,"metrics":["MultiWOZ (Joint Goal Acc)"],"first_row_in_archive_order":{"model":"SGP-DST (base)","paper":"/paper/dialogue-state-tracking-with-a-language-model","metrics":{"MultiWOZ (Joint Goal Acc)":"57.6"},"code_links":[{"title":"chiahsuan156/dst-as-prompting","url":"https://github.com/chiahsuan156/dst-as-prompting"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"},{"leaderboard":"/sota/intent-detection-on-multiwoz-2-2","task":"Intent Detection","dataset_variant":"MULTIWOZ 2.2","rows":1,"metrics":["Accuarcy"],"first_row_in_archive_order":{"model":"MIDAS","paper":"/paper/midas-multi-level-intent-domain-and-slot","metrics":{"Accuarcy":"85.02"},"code_links":[{"title":"adlnlp/Midas","url":"https://github.com/adlnlp/Midas"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"},{"leaderboard":"/sota/slot-filling-on-multiwoz-2-2","task":"Slot Filling","dataset_variant":"MULTIWOZ 2.2","rows":1,"metrics":["F1 score"],"first_row_in_archive_order":{"model":"MIDAS","paper":"/paper/midas-multi-level-intent-domain-and-slot","metrics":{"F1 score":"99.28"},"code_links":[{"title":"adlnlp/Midas","url":"https://github.com/adlnlp/Midas"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"},{"leaderboard":"/sota/task-oriented-dialogue-systems-on-multiwoz-2","task":"Task-Oriented Dialogue Systems","dataset_variant":"MULTIWOZ 2.0","rows":1,"metrics":["BLEU-4","Score"],"first_row_in_archive_order":{"model":"BART (TextBox 2.0)","paper":"/paper/textbox-2-0-a-text-generation-library-with","metrics":{"BLEU-4":"20.17","Score":"100.07"},"code_links":[{"title":"RUCAIBox/TextBox","url":"https://github.com/RUCAIBox/TextBox"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"}],"papers_with_a_benchmark_row":[{"paper":"/paper/midas-multi-level-intent-domain-and-slot","title":"MIDAS: Multi-level Intent, Domain, And Slot Knowledge Distillation for Multi-turn NLU","date":"2024-08-15","rows_on_this_dataset":2,"code_links":1,"syntology":null},{"paper":"/paper/textbox-2-0-a-text-generation-library-with","title":"TextBox 2.0: A Text Generation Library with Pre-trained Language Models","date":"2022-12-26","rows_on_this_dataset":1,"code_links":1,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":1,"samples_ran":1,"samples_unverified":0,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/deepstruct-pretraining-of-language-models-for-1","title":"DeepStruct: Pretraining of Language Models for Structure Prediction","date":"2022-05-21","rows_on_this_dataset":2,"code_links":1,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":13,"samples_ran":7,"samples_unverified":6,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/galaxy-a-generative-pre-trained-model-for","title":"GALAXY: A Generative Pre-trained Model for Task-Oriented Dialog with Semi-Supervised Learning and Explicit Policy Injection","date":"2021-11-29","rows_on_this_dataset":2,"code_links":1,"syntology":null},{"paper":"/paper/dialogue-state-tracking-with-a-language-model","title":"Dialogue State Tracking with a Language Model using Schema-Driven Prompting","date":"2021-09-15","rows_on_this_dataset":4,"code_links":1,"syntology":null},{"paper":"/paper/pretraining-the-noisy-channel-model-for-task","title":"Pretraining the Noisy Channel Model for Task-Oriented Dialogue","date":"2021-03-18","rows_on_this_dataset":1,"code_links":0,"syntology":null},{"paper":"/paper/augpt-dialogue-with-pre-trained-language","title":"AuGPT: Auxiliary Tasks and Data Augmentation for End-To-End Dialogue with Pre-Trained Language Models","date":"2021-02-09","rows_on_this_dataset":2,"code_links":1,"syntology":null},{"paper":"/paper/a-probabilistic-end-to-end-task-oriented","title":"A Probabilistic End-To-End Task-Oriented Dialog Model with Latent Belief States towards Semi-Supervised Learning","date":"2020-09-17","rows_on_this_dataset":1,"code_links":1,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":10,"samples_ran":4,"samples_unverified":6,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/text-to-text-pre-training-for-data-to-text","title":"Text-to-Text Pre-Training for Data-to-Text Tasks","date":"2020-05-21","rows_on_this_dataset":1,"code_links":2,"syntology":null},{"paper":"/paper/soloist-few-shot-task-oriented-dialog-with-a","title":"SOLOIST: Building Task Bots at Scale with Transfer Learning and Machine Teaching","date":"2020-05-11","rows_on_this_dataset":1,"code_links":1,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":3,"samples_ran":1,"samples_unverified":2,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/a-simple-language-model-for-task-oriented","title":"A Simple Language Model for Task-Oriented Dialogue","date":"2020-05-02","rows_on_this_dataset":2,"code_links":1,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":14,"samples_ran":3,"samples_unverified":11,"pointer_only_for_licence":0,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}},{"paper":"/paper/few-shot-natural-language-generation-by","title":"Template Guided Text Generation for Task-Oriented Dialogue","date":"2020-04-30","rows_on_this_dataset":2,"code_links":0,"syntology":null},{"paper":"/paper/few-shot-natural-language-generation-for-task","title":"Few-shot Natural Language Generation for Task-Oriented Dialog","date":"2020-02-27","rows_on_this_dataset":1,"code_links":2,"syntology":null},{"paper":"/paper/task-oriented-dialog-systems-that-consider","title":"Task-Oriented Dialog Systems that Consider Multiple Appropriate Responses under the Same Context","date":"2019-11-24","rows_on_this_dataset":1,"code_links":6,"syntology":null},{"paper":"/paper/semantically-conditioned-dialog-response","title":"Semantically Conditioned Dialog Response Generation via Hierarchical Disentangled Self-Attention","date":"2019-05-30","rows_on_this_dataset":1,"code_links":2,"syntology":{"read_at":"2026-09-24T18:15:14+00:00","samples_harvested":3,"samples_ran":2,"samples_unverified":1,"pointer_only_for_licence":2,"claim":"Per-sample execution on synthesized fixtures; not a correctness claim."}}],"syntology_totals":{"read_at":"2026-09-24T18:15:14+00:00","papers_with_samples":6,"samples_harvested":44,"samples_ran":18,"samples_unverified":26,"pointer_only_for_licence":2,"papers_with_no_sample_that_ran":0,"note":"the per-paper counts above, summed; not a rate"},"papers_note":"The archive never published its papers-using-dataset list; these are papers with a leaderboard row on this dataset's benchmarks."}