Please log in to view the Zoom and Discord sessions' links and to access the individual paper pages.
DSTC11 (09:00-18:40 CEST, Jupiter (1st floor), Watch on Zoom, Chat on Discord)
  • The 11th Dialog System Technology Challenge
    Organizers: Michel Galley, Yun-Nung (Vivian) Chen, Raghav Gupta, Zhang Chen, Paul Crook, Seungwhan (Shane) Moon, Chulaka Gunasekara, Satwik Kottur, Sarik Ghazarian, Behnam Hedayatnia
    In this eleventh challenge, the call for task proposals has resulted into the following 5 different tracks split across two sessions: (1) Ambiguous Candidate Identification and Coreference Resolution for Immersive Multimodal Conversations, (2) Intent Induction from Conversations for Task-Oriented Dialogue, (3) Speech-Aware Dialog Systems Technology Challenge, (4) Robust and Multilingual Automatic Evaluation Metrics for Open-Domain Dialogue Systems and (5) Task-oriented Conversational Modeling with Subjective Knowledge.
    🔗 Workshop schedule
CS4OA (09:00-16:40 CEST, Uranus (1st floor -- updated!), Watch on Zoom, Chat on Discord)
PracticalD2T (09:00-18:00 CEST, Neptune (ground floor -- updated!), Chat on Discord)
YRRSDS 1 (09:30-16:15 CEST, Observatory (9th floor), Chat on Discord)
  • The 19th Young Researchers' Roundtable on Spoken Dialogue Systems
    Organizers: Vojtěch Hudeček, Patrícia Schmidtová, Tanvi Dinkar, Weronika Sieińska, Javier Chiyah-García
    YRRSDS 2023 is an open forum for spoken dialog researchers to discuss their work and research interests. This is the 19th edition of the annual roundtable that provides a networking platform for young researchers in the field. It serves as a playground for stimulating new ideas, sharing tools, and discussing current issues in the Spoken Dialog Systems research.
    🔗 Workshop schedule
Coffee break (10:00-11:00 CEST, 1st floor)
Lunch (12:30-13:30 CEST, Restaurant (1st floor))
Coffee break (15:00-16:00 CEST, 1st floor)
MMNLG (09:00-17:30 CEST, Neptune (ground floor -- updated!), Watch on Zoom, Chat on Discord)
ICARD (09:00-13:00 CEST, Uranus (1st floor), Watch on Zoom, Chat on Discord)
TamingLLM (09:30-18:00 CEST, Jupiter (1st floor -- updated!), Watch on Zoom, Chat on Discord)
  • Taming Large Language Models: Controllability in the era of Interactive Assistants
    Organizers: Robert Xiangru Tang, Devamanyu Hazarika, Di Jin, CHao-Wei Huang, Sherry Tongshuang Wu
    The goal of our workshop is to provide a common ground for experts working on different models and tasks related to controllable text generation. The interplay between research on controlled-generation and instruction-based learning paradigms sits at the heart of our discussions. Here, they can share task-agnostic solutions, engage in evaluations, and discuss task-specific challenges. By encouraging communication, collaboration, and research amongst attendees, the workshop intends to foster insightful debates about the recent progress of the NLG community.
    🔗 Workshop schedule
YRRSDS 2 (09:30-15:00 CEST, Observatory (9th floor), Chat on Discord)
  • The 19th Young Researchers' Roundtable on Spoken Dialogue Systems
    Organizers: Vojtěch Hudeček, Patrícia Schmidtová, Tanvi Dinkar, Weronika Sieińska, Javier Chiyah-García
    YRRSDS 2023 is an open forum for spoken dialog researchers to discuss their work and research interests. This is the 19th edition of the annual roundtable that provides a networking platform for young researchers in the field. It serves as a playground for stimulating new ideas, sharing tools, and discussing current issues in the Spoken Dialog Systems research.
    🔗 Workshop schedule
Coffee break (10:30-11:30 CEST, 1st floor)
Lunch (12:30-13:30 CEST, Restaurant (1st floor))
Coffee break (15:00-16:00 CEST, 1st floor)
Opening (09:00-09:15 CEST, Sun I, Watch on Zoom, Chat on Discord)
Keynote Dupoux (09:15-10:15 CEST, Sun I, Watch on Zoom, Chat on Discord) Chair: Hung-yi Lee
  • Textless NLP: towards language processing from raw audio
    Emmanuel Dupoux
    The oral (or gestural) modality is the most natural channel for human language interactions. Yet, language technology (Natural Language Processing, NLP) is primarily based on the written modality, and requires massive amounts of textual resources for the training of useful language models. As a result, even fundamentally speech-first applications like speech-to-speech translation or spoken assistants like Alexa, or Siri, are constructed in a Frankenstein way, with text as an intermediate representation between the signal and language models. Besides this being inefficient, This has two unfortunate consequences: first, only a small fraction of the world's languages that have massive textual repositories can be addressed by current technology. Second, even for text-rich languages, the oral form mismatches the written form at a variety of levels, including vocabulary and expressions. The oral medium also contains typically unwritten linguistic features like rhythm and intonation (prosody) and rich paralinguistic information (non verbal vocalizations like laughter, cries, clicks, etc, nuances carried through changes in voice qualities) which are therefore inaccessible to language models. But is this a necessity? Could we build language applications directly from the audio stream without using any text? In this talk, we review recent breakthroughs in representation learning and self-supervised techniques which have made it possible to learn latent linguistic units directly from audio which unlock the learning of generative language models without the use of any text. We show that these models can capture heretofore unaddressed nuances of the oral language including in a dialogue context, opening up the possibility of speech-to-speech textless NLP applications. We outline existing technical challenges to achieve this goal, including challenges to build expressive oral language datasets at scale.
Coffee break (10:15-10:45 CEST, Foyer)
INLG Oral Session 1: Trustworthiness of NLG systems (10:45-12:30 CEST, Sun II, Watch on Zoom, Chat on Discord) Chair: Ching-Chi-Chen
Sigdial Oral Session 1: Analysis of discourse and dialogue (10:45-12:30 CEST, Sun I, Watch on Zoom, Chat on Discord) Chair: Gabriel Skantze
Lunch (12:30-13:30 CEST, Restaurant (1st floor))
INLG Oral Session 2: NLG for low-resourced settings (13:30-15:10 CEST, Sun II, Watch on Zoom, Chat on Discord) Chair: Gozde Gul Sahin
Sigdial Poster Session 1 (13:30-15:10 CEST, Foyer, Chat on Discord) Chair: David Schlangen
Coffee break (15:10-15:40 CEST, Foyer)
Sigdial Oral Session 2: LLM for dialogue (15:40-17:00 CEST, Sun I, Watch on Zoom, Chat on Discord) Chair: Dilek Hakkani-Tur
INLG Poster session +flash (15:40-17:00 CEST, Foyer, Chat on Discord) Chair: Ehud Reiter
Keynote Lowe (17:10-18:10 CEST, Sun I, Watch on Zoom, Chat on Discord)
  • Aligning ChatGPT: past, present, and future
    Ryan Lowe
    In this talk I will present different perspectives on the alignment of chatbots like ChatGPT. I’ll review reinforcement learning from human feedback (RLHF), the core training technique behind InstructGPT and ChatGPT, including a brief history of how it was developed. I’ll discuss some of the pitfalls of RLHF, and what is being done today to address them. I’ll then speculate on some of the alignment challenges I expect we’ll face with this new generation of powerful personal assistants, how they could reshape society, and some things we’ll need to do to make sure these changes are good for humans.
Panel (18:10-18:45 CEST, Sun I, Watch on Zoom, Chat on Discord) Chair: David Traum
Welcome reception (19:00-20:00 CEST, Foyer, Chat on Discord)
Keynote di Eugenio (09:00-10:00 CEST, Sun I, Watch on Zoom, Chat on Discord) Chair: Svetalana Stoyanchev
  • Engaging the Patient in Healthcare: Summarization and Interaction
    Barbara Di Eugenio
    Effective and compassionate communication with patients is becoming central to healthcare. I will discuss the results of and lessons learned from three ongoing projects in this space. The first, MyPHA, aims to provide patients with a clear and understandable summary of their hospital stay, which is informed by doctors’ and nurses’ perspectives, and by the strengths and concerns of the patients themselves. The second, SMART-SMS, models health coaching interactions via text exchanges that encourage patients to adopt specific and realistic physical activity goals. The third, HFChat, envisions an always-on-call conversational assistant for heart failure patients, that they can ask for information about lifestyle issues such as food and exercise. All our work is characterized by: large interdisciplinary groups of investigators who bring different perspectives to the research; grounding computational models in ecologically valid data, which is small by its own nature; the need for culturally valid interventions, since our UI Health system predominantly serves underprivileged, minority populations; and the challenges that arise when dealing with the healthcare enterprise.
Coffee break (10:00-10:30 CEST, Foyer)
INLG Oral Session 3: Leveraging Large Language Models for NLG (10:30-12:10 CEST, Sun II, Watch on Zoom, Chat on Discord) Chair: Tom Williams
SIGDIAL Business meeting (12:10-13:00 CEST, Sun I, Watch on Zoom, Chat on Discord) Chair: Dilek Hakkani-Tur
SIGGEN Business meeting (12:10-13:00 CEST, Sun II, Watch on Zoom, Chat on Discord) Chair: Dave Howcroft
Lunch (13:00-14:00 CEST, Restaurant (1st floor))
Sigdial Poster Session 2 (14:00-15:40 CEST, Foyer, Chat on Discord) Chair: Kazunori Komatani
Coffee break (15:40-16:10 CEST, Foyer)
Sigdial Oral Session 4: Language understanding and multimodality (16:10-17:50 CEST, Sun I, Watch on Zoom, Chat on Discord) Chair: Tatsuya Kawahara
INLG genChal presentation, Simon Mille (16:10-17:35 CEST, Sun II, Watch on Zoom, Chat on Discord)
Conference dinner (19:00-22:00 CEST, Klášterní Pivovar Strahov, Chat on Discord)
Virtual Poster Session (09:00-10:00 CEST, Sun I, Chat on Discord) Chair: Kees van Deemter
Keynote Simperl (10:00-11:00 CEST, Sun I, Watch on Zoom, Chat on Discord) Chair: Maria Keet
  • Knowledge graph use cases in natural language generation
    Elena Simperl
    Natural language generation (NLG) makes knowledge graphs (KGs) more accessible. I will present two applications of NLG in this space: in the first one, verbalisations of KG triples feed into downstream KG applications, allowing users with diverse levels of digital literacy to share their knowledge, and contribute to the KG. In the second one, having text representations of KG triples helps us verify the content of a KG against external sources towards more trustworthy KGs. I will present human-in-the-loop solutions to these applications that leverage a range of machine learning techniques to scale to the large, multilingual knowledge graphs modern applications use.
Coffee break (11:00-11:30 CEST, Foyer)
Sigdial Oral Session 5: Topics in open-domain dialogue (11:30-13:10 CEST, Sun I, Watch on Zoom, Chat on Discord) Chair: Mike White
INLG Oral Session 5: NLG for real-world applications (11:30-13:10 CEST, Sun II, Watch on Zoom, Chat on Discord) Chair: Dimitra Gkatzia
Lunch (13:10-14:10 CEST, Restaurant (1st floor))
Sponsors (14:10-14:40 CEST, Sun I, Watch on Zoom, Chat on Discord)
Closing (14:40-15:00 CEST, Sun I, Watch on Zoom, Chat on Discord)
Birds-of-Feather (15:00-15:30 CEST, Sun I, Chat on Discord)
Coffee break (15:00-15:30 CEST, Foyer)