Group recommendation strategies based on collaborative filtering

AUTOR(ES)
DATA DE PUBLICAÇÃO

2003

RESUMO

Nowadays, the amount of information available far exceeds our ability to manage it. We can choose from dozens of TV channels, thousands of movies, millions of books, billions of on-line documents. When we have to make choices without full knowledge of the alternatives, a common approach is to rely on recommendations of trusted persons. In the 1990s computer recommender systems have appeared to automatize the recommendation process. Today, popular sites like Amazon.com give thousands of recommendations every day. However, while many activities are carried out in groups, like going to the movies with friends, current systems focus only on recommending for sole users. This brings out the need of systems capable of performing recommendations for groups of people, a domain that has received little attention in the literature. In this work, we investigate the problem of generating automatic group recommendations, making connections with problems considered in other research areas like social choice and social psychology. We propose two methods based on collaborative filtering to generate recommendations: one that aggregates individual recommendations based on an existing technique of classification of alternatives which uses fuzzy majority; and a novel methodology that builds a model for the group using techniques from symbolic data analysis. Finally, we empirically evaluate the proposed methods to see their behavior for groups of different sizes and degrees of homogeneity. To this end, we develop an evaluation framework that quantifies the quality of the group recommendations based on a set of metrics that reflect desirable properties these recommendations should have

ASSUNTO(S)

information filtering collaborative filtering recommendations for groups symbolic data analysis ciencia da computacao recommender systems

ACESSO AO ARTIGO

Documentos Relacionados