DialoKG: Knowledge-Structure Aware Task-Oriented Dialogue Generation

Research output: Contributions to collected editions/worksArticle in conference proceedingsResearchpeer-review

Authors

Task-oriented dialogue generation is challenging since the underlying knowledge is often dynamic and effectively incorporating knowledge into the learning process is hard. It is particularly challenging to generate both humanlike and informative responses in this setting. Recent research primarily focused on various knowledge distillation methods where the underlying relationship between the facts in a knowledge base is not effectively captured. In this paper, we go one step further and demonstrate how the structural information of a knowledge graph can improve the system's inference capabilities. Specifically, we propose DialoKG, a novel task-oriented dialogue system that effectively incorporates knowledge into a language model. Our proposed system views relational knowledge as a knowledge graph and introduces (1) a structure-aware knowledge embedding technique, and (2) a knowledge graph-weighted attention masking strategy to facilitate the system selecting relevant information during the dialogue generation. An empirical evaluation demonstrates the effectiveness of DialoKG over state-of-theart methods on several standard benchmark datasets.

Original languageEnglish
Title of host publicationFindings of the Association for Computational Linguistics : NAACL 2022 - Findings
Number of pages15
PublisherAssociation for Computational Linguistics (ACL)
Publication date01.01.2022
Pages2557-2571
ISBN (electronic)9781955917766
DOIs
Publication statusPublished - 01.01.2022
Externally publishedYes
Event2022 Findings of the Association for Computational Linguistics - NAACL 2022 - Online + Hyatt Regency Seattle, Seattle, United States
Duration: 10.07.202215.07.2022
https://2022.naacl.org/downloads/handbook-final-v2.pdf
https://aclanthology.org/2022.findings-naacl.0/

Bibliographical note

Publisher Copyright:
© Findings of the Association for Computational Linguistics: NAACL 2022 - Findings.

Recently viewed

Publications

  1. Do Serious Breaches Give Rise to Any Specific Obligations of the Responsible State?
  2. Hypothesenprüfung, Theorieentwicklung und Erkenntnisfortschritt in der Mathematikdidaktik
  3. Die Vitamine liegen unter der Schale
  4. Paul Paolucci. Marx’s Experiments and Microscopes: Modes of Production, Religion, and the Method of Successive Abstractions
  5. Positive Empathy and Prosocial Behavior
  6. Newspapers and the circulation of academic knowledge
  7. Transnational migration and the emergence of the European border regime
  8. Zukunft Publikum
  9. Correlation of six anthropogenic markers in wastewater, surface water, bank filtrate, and soil aquifer treatment
  10. Erratum zu
  11. The Revolution Will Not Be Liked
  12. APPschaffung der Lehrkraft?
  13. Stakeholder signalling and strategic niche management
  14. Artificial intelligence, systemic risks, and sustainability
  15. Preferences and policy - Consuming art and culture in Baltimore and Hamburg
  16. Léger, Fernand
  17. Arbeitskampf
  18. Software gibt es doch
  19. Ökologieorientiertes Management
  20. Videoüberwachung am Arbeitsplatz — eine Grenzziehung
  21. Overhead Projector
  22. Integrated Reporting zwingt die Unternehmen zu einer konsequenten Reorganisation der Berichtsstrukturen
  23. Guest editorial
  24. International scaling of sustainability continuing professional development for in-service teachers
  25. Clonal expansions of pathogenic CD8+ effector cells in the CNS of myelin mutant mice.
  26. Simulating the fate and release of metazachlor and its transformation products metazachlor oxalic acid and sulfonic acid in a small-scale agricultural catchment
  27. The Collaborative Production of Meaningful Measure(ment)s