Title: Constraint-based feature selection for location intelligent data mining.
Abstract:
Recommender Systems (RS) are software systems which, are designed to
aid the users. They are based on the observation that, in the real
world, people heavily rely on the recommendations of others when making
daily decisions. Such decisions are, for example,
items to buy, what movies to see, and so forth. Recommender
Systems thus involve the design of algorithms which seek to mimic this
behavior. Recommender systems may produce non-personalized
recommendations, such as best sellers’ lists; or personalized
recommendations, based on person’s likes, preferences and behaviors.
The new trend of system development is the creation of Recommender
Systems which provide personalized services to each user.
Unfortunately, the current approaches base their results on a single
criterion or pattern, and require a large number of user’s data to
generate accurate recommendations. Our research focuses on the creation
of a multiple criteria RS, which is tailored for a specific user at a
specific location. Our algorithm creates a personalized preference
model for each user, based on location aware (user current location and
level of mobility), special needs, and user profile. User profiles are
generalized from a large scale demographic database. We describe our
algorithm and show a case study where we applied our methodology within
the PeRS (Personal Recommender System) environment to recommend
“events” (festivals/fairs, exhibitions, etc), in the Ottawa region.