Sentiment and preference guided social recommendation

© Springer International Publishing Switzerland 2014. Social recommender systems harness knowledge from social experiences, expertise and interactions. In this paper we focus on two such knowledge sources: sentiment-rich user generated reviews; and preferences from purchase summary statistics. We fo...

ver descrição completa

Detalhes bibliográficos
Autores: Chen, Yokeyie, Ferrer, Xavier, Wiratunga, Nirmalie, Plaza, Enric
Formato: artículo
Estado:Versión aceptada para publicación
Fecha de publicación:2014
País:España
Recursos:Consejo Superior de Investigaciones Científicas (CSIC)
Repositorio:DIGITAL.CSIC. Repositorio Institucional del CSIC
OAI Identifier:oai:digital.csic.es:10261/131950
Acesso em linha:http://hdl.handle.net/10261/131950
Access Level:acceso abierto
Palavra-chave:Sentiment analysis
Preference graph
Aspect extraction
Social recommender systems
Descrição
Resumo:© Springer International Publishing Switzerland 2014. Social recommender systems harness knowledge from social experiences, expertise and interactions. In this paper we focus on two such knowledge sources: sentiment-rich user generated reviews; and preferences from purchase summary statistics. We formalise the integration of these knowledge sources by mixing a novel aspect-based sentiment ranking with a preference ranking. We demonstrate the utility of our proposed formalism by conducting a comparative analysis on data extracted from Amazon.com. In particular we show that the performance of the proposed aspect based sentiment analysis algorithm is superior to existing aspect extraction algorithms and that combining this with preference knowledge leads to better recommendations.