| Data Mining is a large research area, with a lot of open problems. We propose to consider data mining according to a general approach of "Experience Management from CHI Traces". Our domain involves collective usages of NTIC information systems. By collective usages, we mean the usage issued from communities such as citizens, handicapped persons, tourists, a company's employees, members of a scientific community, etc.
The objectives of proposed researches aim to make some propositions of efficient methods 1) to extract knowledge/experience from large databases (such logs) in order to be used as indexing methods inside an experience management framework; and 2) to update the experience memory. Among planned data mining technics, we aim mainly those extracting classification structures, in particular neural or ascendant hierarchical approaches for classifying user sessions.
The targeted user group will be chosen in the first step in relation with our industrial partners (Usage laboratory, MobIVIP Predit project) or local collectivities or associations (Antibes City, CASA, Travisa). Then a moke up of a recommender system inside an existing web-based information system will be developed in order to validate such propositions. Such a recommender based on reusing user interactions will be developed with CBR*Tools our CBR framework in Java.
AxIS: URL: http://www-sop.inria.fr/axis/
|
