Information Technology Reference
In-Depth Information
IFWeb [ASN 97] carries out the search and filtering of documents by taking into
account the specific needs of the user; when a document is selected by a user, the
system searches the web for similar documents and shows them to the user, ranking
them in order of relevance.
IFM (
Intelligent File Manipulator
) [VIR 02] is a system that aids the graphic
handling of file systems; it intervenes automatically and suggests advice when a user
makes inconsistent approaches or commits handling errors. IFM bases itself on
stereotypes and includes a mechanism for recognizing the goals of the user.
InfoSleuth [NOD 00] relates the queries of the user to the corresponding service
providers. The work of InfoSleuth agents is based on the use of ontologies, which
enable them to specify the queries or to break them down, and then to fuse collected
information together based on heterogeneous and distributed sources.
WebMate [KEE 00] is an agent for aiding information search on the web. It
learns the profile of the user, prepares personalized news for him and helps him
improve his information search. The profile of a user is an ensemble of keywords
found in the pages that he has selected. These pages are positive examples of the
machine learning algorithm; this algorithm uses the frequency of appearances of
words in the pages. There is a mechanism of keyword “expansion”: the addition of
words with a similar connotation to a given keyword in order to specify its meaning.
Stemming from previous works at the Laboratoire d'Automatique, de Mécanique
et d'Informatique industrielles et Humaines (LAMIH), MAPIS (MultiAgent
Personalized Information System) [PET 03a], [PET 04], [PET 06] is a Web
application based on software agents that help users of public people transport with
their itinerary choice. The profile of the user corresponds to the associated
weighting given in relation to the different modes (bus, train, metro, walking, etc.),
the length of the journey, the number of changes and the cost of the journey. MAPIS
uses a reinforcement learning mechanism for management of the user profile.
Other systems are based on the profile of the user to provide personalized
information in specific domains, such as
Gulliver's Genie
[OHA 03] for tourism and
Smart Radio [HAY 04] for music. The former uses localization, direction and user
preferences to search for cultural or tourist places that could be of interest to him,
and sends this personalized information on a PDA. The system is based on the
Beliefs, Desires and Intentions
or BDI model [RAO 91] to infer the mental state of
the user. The latter enables users to listen to their favorite music or to a musical
program recommended by the system. In order to know their musical tastes, Smart
Radio asks its users to rate an ensemble of musical pieces, or musical programs,
from one to five (five being the highest mark), in order to establish user profiles.
