Recent progress in the field of NLP impacts all of its subfields and extends its domain of applications. Central in these developments is automated dialogue, either with chatbots (scripted, conversational, cognitive) or personal assistants, ubiquitous as services via phones or commercial websites.
At the same time, new written media continue to grow and most of them involve some sort of interaction: chats, forum, emails, micro-blogging, or collaborative instant messaging services.
This progress and the generalization of natural language for interaction also make way for novel approaches taking into account multiple modalities (images, video) and the situation in which dialogues take place. In this context, all aspects of conversation analysis and dialogue system development are concerned, whether communication is oral or written, task-oriented or open.
The prevalence of neural network based approaches in NLP further influences approaches to dialogue, putting into focus different issues, such as automatic generation of diverse and natural responses, although these approaches sometimes minimize the role of understanding and take into account linguistic and extra-linguistic context in a very limited way.
New approaches also bring new problems that pervade machine learning in general : black box models are not easy to explain, and blurring the line between human and machines generates ethical side-effects, such as conversational agents reproducing biases and prejudices present in their training data. Paucity of supervised data is another issue for machine learning approaches, which motivates collection of auxiliary data, semi-automatically annotated data or artificial data far from real use cases. This can generate a gap between theory and practice that needs to be addressed.
For this special issue, the TAL journal invites contributions on all aspects related to the analysis of written or transcribed conversations, to the development and evaluation of dialogue systems, to data collection for all interaction modalities, and to ethical and social issues pertaining to dialogue and its applications.
Topics
- Conversation Analysis
- speaker recognition
- pragmatic phenomena in conversation
- thread extraction and analysis
- speech acts
- dialogue structure
- disfluencies
- Dialog systems
- task oriented dialogues
- open-domain conversation systems
- dialogue management and state tracking
- planning
- response generation
- intention detection
- Data and tools
- data collection, corpus creation
- evaluation methodologies
- Multimodality
- dialogue related to visual tasks
- situated communication
- sign-language conversation
- Social and ethical issues :
- bias in systems and chatbots
- detecting automated systems
- Applications
- personal assistants
- meeting assistants
- chatbots/technical assistance
- case studies of interaction analysis
Invited editor for this issue: Liesbeth Degand
Journal editor for this issue: Philippe Muller
IMPORTANT DATES
- Deadline for submission: 20th of June 2020 27th of June 2020
- First round of review notifications: 15th of September 2020
- Second round notifications: 15th of December 2020
- Publication on line: January 2021