Post-Doctoral Research Visit F/M Automatic analysis of alignment between speakers in conversations
Société : INRIA Lieu : Paris (Île-de-France)
Descriptif du poste
Société : INRIA Catégorie : Offre emploi CDD Filiere : IT/Etudes, développement et intégration Lieu : Paris (Île-de-France)
Mission
Contexte et atouts du poste
This postdoc position is offered in the context of the ANR PRME SINNet project (Socio-Inspired Neural Networks for Conversational Systems), aiming to develop socially-aware neural conversational models which are able to adapt their language and behavior to the agent-user social relationship.
One of the key human behaviors that is desirable for automatic systems to imitate is alignment, also called entrainment, which consists in mirrorring the interlocutor's behavior and language use. This postdoc position will specifically be devoted to the characterization and modeling of alignment at the level of word meaning.
Mission confiée
The postdoc will focus on cases of miscommunication due to a lack of lexico-semantic alignment between speakers in a conversation.
The goals will be to:
Investigate, analyze and characterize this phenomenon in human interaction: how humans produce, signal, and solve misunderstandings or disagreements due to specific lexical elements in dialog;
Produce a semi-automatically annotated corpus with instances of this phenomenon;
Propose models that can identify word usages that are potential triggers or causes of miscommunication, as well as pinning down the reason why they may be problematic and for whom;
Build or adapt systems that are able to both avoid producing potential triggers, adapting to the speaker and the conversational situation; mirror human behavior and repair strategies when faced with such triggers, e.g. asking for disambiguation.
Provide ways of evaluating the proposed methodology.
Principales activités
The successful postdoc candidate will be part of INRIA's ALMAnaCH team, specialized in NLP. Besides the research tasks mentioned under Assignments, they will be expected to participate and collaborate in the team's activities, including seminars and reading groups.
Skills
PhD in Natural Language Processing, Computational Linguistics, Computer Science with a specialization in Machine Learning, or related fields;
Knowledge of deep learning as well as more traditional machine learning techniques;
Desired prior experience on conversational AI, word meaning representations and corpus annotation;
Strong programming skills (Python);
Fluency in English.
Avantages
Subsidized meals
Partial reimbursement of public transport costs
Leave: 7 weeks of annual leave + 10 extra days off due to RTT (statutory reduction in working hours) + possibility of exceptional leave (sick children, moving home, etc.)
Possibility of teleworking
Flexible organization of working hours (after 12 months of employment)
Professional equipment available (videoconferencing, loan of computer equipment, etc.)
