Recommendations Using Information from Selected Sources with the ISIRES Methodology

SILVANA ACIAR; JOSEFINA LÓPEZ HERRERA; JOSEP LLUIS DE LA ROSA I ESTEVA

Frontiers in Artificial Intelligence and Applications. AI Research & Development

Ed. IOS Press

Lugar: Amsterdam, The Netherlands; Año: 2008 vol. 146 p. 258 - 265

0922-6389

Recommender systems have traditionally made use of the all variety of sources to obtain the suitable information to make recommendations. There are costs associated with the use of information sources those costs are an important determinant in the choice of which information sources are finally used. For example recommendation can be better if the recommender knows where is the suitable information to predict user´s preferences to offer products. Sources that provide in-formation that is timely, accurate and relevant are expected to be used more often than sources that provide irrelevant information. This paper shows how the precision of the recommendations using either Collaborative Filtering (CF) or Content-Base Filtering (CBF) increases by selecting the most relevant information sources based on their intrinsic characteristics.