Research in the Department of Databases and Machine learning is centered around statistical and symbolic machine learning and distributed databases. Probabilistic methods, neural networks, fuzzy logic, aggregation, explanation-based learning techniques, data replication, distributed query and transaction processing are among the methods currently used for fundamental and applied research.

The target applications are situated in a large variety of domains which range from adaptive or selective information retrieval in text, web, images and videos, to efficient data management in large scale networks (P2P, grid, PC cluster).

Content-based information retrieval and database querying are two complementary approaches for accessing heterogeneous data. Methods are proposed for the efficient processing of structured or semi-structured information. For instance, XML documents are automatically organized by means of Bayesian networks, specific information is filtered. Efficiency is obtained by using advanced database techniques including data replication, load balancing, logical clustering indexing and adaptive routing. Multimedia information retrieval is based on semantic indexation and enhanced through the use of different medias.

Risk analysis, heterogeneous data fusion, knowledge discovery, are particularly studied in large databases. Web usage mining, prefetching, recommendation systems, interface customization help the user to interact with web pages and to extract relevant information. More generally, research concerns user modllig and profiling, adaptive hypermedia and interface personalization, for instance content adaptive navigation interfaces and pen-based interfaces.