480 Int. J. Business Information Systems, Vol. 14, No. 4, 2013
Copyright © 2013 Inderscience Enterprises Ltd.
An effective recommendation based on user
behaviour: a hybrid of sequential pattern of user and
attributes of product
Mojtaba Salehi
Industrial Engineering Department,
K.N. Toosi University of Technology,
1999143344, Tehran, Iran
Fax: +982188674858
E-mail: m_salehi61@yahoo.com
E-mail: m.salehi@kntu.ac.ir
Abstract: Recommender system is a promising technology for companies to
present personalised offers to their customers. But this technology suffers from
sparsity problem. In addition, most researches are based on explicit rating. But
most users do not spend time for rating of products. Therefore, this research
proposes an effective recommendation based on user behaviour. Since users
express their opinions implicitly based on some specific attributes of products,
we introduce a preference matrix that can collect user preferences based on
attributes of products. In addition, since there are some sequential patterns in
purchasing of products, we use weighted association rules to discover these
patterns to improve the quality of recommendation. The method outperforms
current algorithms and alleviates sparsity problem. Main contribution is
implementation of a user behaviour-based recommendation method that
discovers interest of users based on implicit rating of product attributes. In
addition, this approach uses sequential pattern of purchasing to improve the
quality of recommendation.
Keywords: personalisation; content-based filtering; CBF; personalised
recommender; sparsity; product attributes; attribute-based; sequential-based;
association rules; information overload; collaborative filtering.
Reference to this paper should be made as follows: Salehi, M. (2013)
‘An effective recommendation based on user behaviour: a hybrid of sequential
pattern of user and attributes of product’, Int. J. Business Information Systems,
Vol. 14, No. 4, pp.480–496.
Biographical notes: Mojtaba Salehi received his BSc degree from Shahid
Bahonar University in 2004 and MSc degree from Tehran University in 2006
and his PhD degree from Tarbiat Modares University in 2013. During 2012, he
was a Researcher in TU/e, Mathematics and Computer Science Department. He
is currently working as an Assistant Professor of K.N. Toosi University of
Technology (KNUT). His research areas of interest include soft computing,
data mining, combinatorial optimisation, recommender systems and applied
multivariate analysis.