Description
Date depot: 17 septembre 2019
Titre: User Models for Solving Complex Information Needs
Directeur de thèse:
Benjamin PIWOWARSKI (ISIR (EDITE))
Directeur de thèse:
Sylvain LAMPRIER (LERIA)
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Sciences de l’information et sciences du vivant
Resumé:
The Ph.D. will mainly deal with learning representations of complex search tasks. By analogy with the importance of query and document representation in traditional IR models, this step is fundamental for designing task-based information access models. Recent work tackled the problem of task representation from the perspective of discovering coherent successive queries in search sessions. Our perspective is radically different since we attempt to build the representations of tasks that support their completion based on system-driven assistance. In this line of work, we plan to follow a deep learning approach where subtasks are embedded in probabilistic representation spaces, allowing to model both the evolution dynamics of the search toward the final task completion and the subtasks that have to be achieved before then. Recent advances in stochastic representation (e.g., variational auto-encoders, generative adversarial networks, stochastic recurrent networks) will guide our research directions. The aim is to encode the relationships between subtasks in a latent space that can be used to design complex task-driven IR models.
Doctorant.e: Mustar Agnès