BEGIN:VCALENDAR VERSION:2.0 PRODID:-//wp-events-plugin.com//7.4.5//EN TZID:Europe/Paris X-WR-TIMEZONE:Europe/Paris BEGIN:VEVENT UID:0-324@lisn.upsaclay.fr DTSTART;TZID=Europe/Paris:20251008T093000 DTEND;TZID=Europe/Paris:20251008T093000 DTSTAMP:20251006T134245Z URL:https://www.lisn.upsaclay.fr/evenements/vers-des-dialogues-plus-nature ls-integrer-des-competences-de-dialogue-informel-a-des-agents-conversation nels-orientes-tache/ SUMMARY:Towards More Natural Dialogues: Integrating Chitchat Capabilities i nto Task-oriented Dialogue Agents DESCRIPTION:Rapporteurs\nFrançois Portet\, Professeur\, Université Grenob le-Alpes\nYves Lepage\, Professeur\, Waseda University\nExaminateurs\nSakr iani Sakti\, Professeure\, Nara Institute of Science and Technology (NAIST )\nSophie Rosset\, Directrice de Recherche CNRS\, Université Paris-Saclay \nOndřej Dušek\, Assistant Professor\, Charles University\, Prague\nAbs tract\nThe ability to hold natural\, human-like conversations has long bee n regarded as a hallmark of intelligence. Contemporary dialogue systems ar e typically divided into two categories: task-oriented dialogue (TOD)\, fo cused on efficiency and goal completion\, and open-domain chitchat\, cente red on engagement and social connection. Yet human conversations rarely co nform to this dichotomy. Everyday dialogue interleaves transactional and s ocial goals\, suggesting that dialogue systems should likewise integrate b oth modes to achieve more natural interactions.\nThis thesis investigates the integration of chitchat into task-oriented dialogue\, advancing the ar gument that these modes are not competing but complementary. The work is s tructured in two parts. The first part provides a state-of-the-art overvie w of dialogue systems\, tracing the historical evolution of TOD and chitch at agents\, the datasets and evaluation paradigms that shaped them\, and t he emergence of integrated “inter-mode” approaches. It also introduces relevant natural language processing and machine learning foundations\, p ositioning recent advances in large language models as critical enablers o f hybrid dialogue systems.\nThe second part presents original research con tributions that explore how chitchat can enhance task-oriented dialogue in practice. Four studies are reported. First\, a comparative study evaluate s different strategies for injecting chitchat into TOD\, revealing distinc t lexical patterns and showing how certain forms of social talk improve di versity and engagement. Second\, a unified modeling approach demonstrates that incorporating user emotion detection into an end-to-end TOD system im proves both task success and conversational naturalness\, underscoring the value of affective signals in practical interactions. Third\, we augment TODs with synthetic user backstories\, simulating inter-mode scenarios by embedding elements of a user narrative into task requests. This training d ata supports both the evaluation and improvement of model robustness to su ch common inter-mode inputs. Finally\, a few-shot prompting approach shows that large language models can flexibly combine task-oriented and chitcha t capabilities without architectural modifications\, highlighting the prac ticality of LLM-based methods for building unified dialogue agents.\nToget her\, these contributions advance the development of inter-mode dialogue s ystems by addressing key challenges such as dataset limitations\, unified modeling of social and task signals\, and robustness to blended user input . The thesis demonstrates that chitchat and TOD can be coherently integrat ed without necessarily sacrificing task performance. Ultimately\, it argue s that moving beyond the artificial separation of TOD and chitchat is esse ntial for creating dialogue systems that are not only effective and reliab le\, but also more natural\, empathetic\, and enjoyable.\nLien visio conf érence : ici CATEGORIES:STL,Thèses et HDR LOCATION:LISN Site Belvédère\, Rue du Belvédère Campus universitaire b ât 507 91405 Orsay\, France X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=Rue du Belvédère Campus universitaire bât 507 91405 Orsay\, France;X-APPLE-RADIUS=100;X-TITLE=LIS N Site Belvédère:geo:0,0 END:VEVENT BEGIN:VTIMEZONE TZID:Europe/Paris X-LIC-LOCATION:Europe/Paris BEGIN:DAYLIGHT DTSTART:20250330T030000 TZOFFSETFROM:+0100 TZOFFSETTO:+0200 TZNAME:CEST END:DAYLIGHT END:VTIMEZONE END:VCALENDAR