The omnipresence of data processing and mobile telephony in our life (computers, PDA, GSM, GPS...) along with the evolution of wireless technologies opens the door towards new habits. To avoid being submerged by too much information it is necessary to equip each electronic component present in user's daily life with capacities to take into account his needs according to his actions, to assist him while learning and anticipating on his behavior in the most autonomous way. Personalization is clearly situated in this objective; it enables a user profile construction which has to dynamically evolve. It also has to take into account new preferences, needs and interests of this user and to forget old ones. This paper proposes a local, cooperative and real-time multi-agent approach to build adaptive and incremental profiles. First, documents are sequentially parsed, which leads to the construction of a Temporary Terminological Network (TTN). This Network is then merged with other document's extracted networks, in order to create a Permanent Terminological Network (PTN), relevant to the studied collection and used to index this collection thanks to a clustering approach. Preliminary results of the built system are then presented as well as perspectives.
展开▼