Conversational agents are becoming widely used for various tasks, domains and settings (e.g., personal assistants, shopping assistants, customer service agents, personal tutors). This increasing usage is driven by recent advances in artificial intelligence and natural language processing along with increasingly capable chat development environments, leading to improvements in conversational richness and robustness.
One of the reasons that conversational agents are embraced as an effective form of interaction is their ability to dynamically generate content and adapt to individual users. However, personified natural language interfaces often lead to high user expectations on sensing and adaptive capabilities, not only to satisfy individual information needs, but also to behave in socially appropriate or favorable ways.
Therefore, conversational agent systems present an extremely rich and challenging research space to address issues in user awareness and adaptation in many dimensions such as user profiles, contexts, personalities, emotions, conversational style, social dynamics, etc. Nowadays, the implementation of such systems often involves deep learning, reinforcement learning, active learning, and other machine learning approaches. It is important to consider how these interaction aspects should be handled by algorithms and system components such as language understanding, dialog management, and language generation, or in a setting of an end-to-end conversation modeling.
Adaptive interfaces have been a long-standing interest in the human-computer interaction (HCI) community, which often takes extensive effort in studying users, prototyping, and evaluating to design adaptive actions of computing systems. However, this effort is sometimes isolated from the challenges in developing the sensing capabilities of systems, and the opportunities in leveraging data-driven approaches and computational intelligence.
The purpose of this workshop is to bring together researchers in AI, NLP, user modeling, and HCI communities from both industry and academia. Through a focused and open exchange of ideas and discussions, we will work to identify central research topics in user-aware conversational agents and develop a strong interdisciplinary foundation to address them.
This will be a one-day workshop including paper, demo, and poster presentations, invited talks, and discussion sessions. Invited talks will feature leading experts on the topic from both HCI and AI perspectives. Discussion sessions will focus on enumerating key challenges in user-aware conversational agents and developing an interdisciplinary research agenda.
Passionate in technology and art, Rossana has a multifaceted background mixing computer science and interaction design. She aims at combining different disciplines to ehnance the way of leading projects on complex systems, bridging industrial and academic teams and realizing publications and patents on important topics about Interactive Intelligent Systems.She increased her technical and soft skills coordinating R&D projects in the Telecommunications context. In recent years she has been responsible for innovation and digital transformation projects leveraging her previous experience as researcher and adding up to the technical expertise and relational skills, a touch of extra determination, so necessary to nail down successfully ambitious innovative projects in big companies. She is proud to be a member of the TIM Innovation Department and to have the opportunity to continuously share and enhance her experience teaching HCI and ID at the University of Torino.
09:00 - 09:15 AMWelcome
09:15 - 10:15 AMNovel tasks
Dynamic Natural Language User Interfaces Using Microservices
Riyadh Mahmood, Arth Joshi, Adwait Lele, and Jay Pennington
Assessing Language Learners’ Free Productive Vocabulary with Hidden-task-oriented Dialogue Systems
Dolça Tellols, Takenobu Tokunaga, and Hilofumi Yamamoto
10:30 - 11:00 AMCoffee-Break
11:00 - 12:00 PMKeynote:
User awareness in TIM Conversational Agent & Platform: combining human centered design and data driven evolution
Rossana Simeoni, TIM
12:00 - 12:30 PMEvaluation
How does your Alexa behave? Evaluating Voice Applications by Design Guidelines Using an Automatic Voice Crawler.
Xu Han and Tom Yeh
12:30 - 14:00 PMLunch
14:00 - 15:30 PMPersonalization and adaptation
A Snooze-less User-Aware Notification System for Proactive Conversational Agents
Yara Rizk, Vatche Isahagian, Merve Unuvar, and Yasaman Khazaeni
A Multi-Turn Emotionally Engaging Dialog Model.
Yubo Xie, Ekaterina Svikhnushina, and Pearl Pu.
A Virtual Customer Assistant for the Wealth Management domain in the UWMP project.
Andrea Iovine, Marco de Gemmis, Fedelucio Narducci, Giovanni Semeraro, Da-vide Ingoglia Doriana Filisetti, and Georgios Lekkas
15:30 - 16:00 PMCoffee-Break
16:00 - 17:00 PMOpen Discussion and Networking
The workshop solicits submissions in all aspects of user-aware and adaptive conversational agents including:
- User modeling for conversational agents and multi-modal interactions
- Sensing capabilities of agents (e.g., emotion, personality, contexts, social dynamic, etc.)
- Agent adaptation through language generation, dialog management and conversation modeling
- Personalization and adaptation algorithms inspired by behavioral or psychological theories
- Adapting agent interactions for user engagement
- Transparency and control of adaptive agents
- Novel methods for evaluating adaptive agents
- User interactions with and perceptions of adaptive agents
- Case studies of adaptive agents for different uses cases (e.g. collaborative tasks, decision support, social agent) and different domains (e.g. healthcare, finance, education)
user2agent encourages original and relevant work of two publication types:
- full papers (up to 10 pages), including work in progress, perspective papers, and lessons learned. They will be presented either as oral presentations or posters.
- position papers (up to 4 pages), which will be presented as posters with a possibility to be accompanied by a demo.
- All papers should follow ACM SIGCHI templates: https://sigchi.org/templates/, and submitted electronically as a single PDF file through the EasyChair submission system: https://easychair.org/conferences/?conf=user2agent0
- All submissions will undergo a peer-review process. Reviewers will consider originality, significance, technical soundness, clarity, and relevance to the workshop’s topics. The reviewing process will be double-blind.
- Ron Artstein,
USC Institute for Creative Technologies, Playa Vista, California, USA
- Michal Shmueli-Scheuer,
IBM Research AI, Haifa, Israel
- Hao Fang,
Microsoft Semantic Machines, Redmond, WA, USA
- Yasaman Khazaeni,
IBM Research AI, Cambridge, USA
- Q. Vera Liao,
IBM Research AI, TJ Watson Research Center, USA
Program Committee (confirmed)
- Jonathan Herzig, Tel-Aviv University
- Maxine Eskenazi, CMU
- Ramesh Manuvinakurike, USC
- Kallirroi Georgila, USC
- Erik T. Mueller, Capital One
- Oren Sar Shalom, Intuit
- Ali Kebarighotbi, Amazon
- Werner Geyer, IBM
- Stefan Ultes, University of Cambridge, UK
- Payal Bajaj, Microsoft, USA
- Kevin Bowden, University of California Santa Cruz, California, USA
- Shereen Oraby, Amazon
- David Konopnicki, IBM
- Stefan Kopp, Bielefeld University, Germany
- Ziming Huang, IBM Research
- Jie Ma, IBM Research
|Submission deadline:||27 December 2019|
|Notifications to authors:||14 January 2020|
|Camera-ready of accepted papers:||17 January 2020|
|Workshop date:||17 March 2020|