This site describes the activites of the MRIM research group.
The MRIM group studies methods for satisfying user information needs. These needs have largely evolved over time, as well as contexts in which they are formulated and the methods by which they are satisfied.
The context of Information Retrieval (IR) has broadened not only to include all types of data (texts, images, video, etc.), but also to include all types of users, and IR has been widely used through other applications (filtering, recommending systems, social networks, etc.). More recently, concerns about the fairness and the transparency of IR systems has emerged and the recent GDPR European Union regulation has introduced a “right to explanation” on algorithmic decision-making, which all IR systems or methods are subject to. This is especially challenging, as most of them now involve “deep learning black boxes”.
As in many other domains, learning-based methods, with, as a corollary, the use and building of large annotated collections, have become dominant for almost all aspects of IR. Such point should not however be considered as exclusive, as other fundamental aspects of information retrieval (like new extensions of existing models of IR) are still under concern in this research field.
MRIM has conducted research on the following themes:
• Modeling of IR systems
• Semantic indexing
• Multimedia Indexing
• Transparency and explainability
• Social Networks and Personalization
• IR in Under-Resourced Languages
• Mobile and Cultural Heritage
• Medical IR
• IR for E-tourism
• IR and Smart Cities
• Evaluation of IR systems
All our research work is described here.
Our 2019 activity report is available here.