Goal-oriented conversational systems can unlock a vast amount of information
in the associated documents, in which written and visual content dominate the
way that individuals and organizations choose to present their interests and
knowledge to the world. We posit that one scalable way to create personalized
conversational systems is to have them arise directly from such content.
There are significant individual research threads that show promises in
handling heterogeneous knowledge embedded in the documents, including (1)
unstructured content such as text passages (CoQA, QuAC, ShARC, DoQA); (2)
semi-structured content such as tables or lists (SQA, HybridQA); (3)
multimedia such as images and videos with associated textual descriptions
(RecipeQA, PsTuts-VQA); (4) or structured data specified by schema such as
RDFa or Microdata in the webpages. The purpose of this workshop on
Document-grounded Dialogue and Conversational QA is to invite researchers to
bring their individual perspectives on the subject, and to advance the field
in joint effort by providing a shared task and competition.
**Website**
https://doc2dial.github.io/workshop2021/
**Call for Papers**
We welcome submissions of original work as long or short papers as well as
non-archival papers. All accepted papers will be presented at the workshop.
We encourage submissions from a wide range of topics, including but not
limited to:
- Document-grounded dialogue systems
- Conversational machine reading comprehension
- Machine reading comprehension / contextual QA / sequential QA over
documents
- Conversational search among domain documents
- Topical chat based on associated documents
- Parsing semi-structured document content for sequential QA or dialogue /
table reading
- Policy learning for document-guided dialogue agents
- Task-oriented knowledge-grounded multimodal dialogue
-- task-oriented situated dialogue;
-- task-oriented domain-knowledge-grounded VQA;
-- task-oriented video-based dialogue;
- Evaluations for document-grounded dialogue
- Transfer learning for document-grounded dialogue systems on low-resource
domains
- Summarization of document-grounded dialogues
- Model interpretation of document-grounded dialogues
**Important Dates**
Workshop Paper Due Date: May 6, 2021 (AoE)
Notification of Acceptance: June 1, 2021 (AoE)
Camera-ready papers due: June 7, 2021 (AoE)
Workshop Dates: August 5 or 6, 2021