Affiliations: Department of Computing, The Hong Kong Polytechnic University, Hung Hom, Hong Kong, China
Correspondence:
[*]
Corresponding author: Anming Li, Department of Computing, The Hong Kong Polytechnic University, Hung Hom, Hong Kong, China. Tel.: +852 5345 4007; E-mail: lianming03@gmail.com
Abstract: We concern the issue of preference recommendation towards more reliable recommender system. General recommender system provides a collection of items or several bests of them, on the basis of a fixed preference constraint. However, the realistic preference from users may be such complicated that makes conventional recommender system unreliable. In real-world applications, e.g. travel planning or hotel selection, specific constraints may be involved like a limited travel time or an appropriate budget for hotel accommodation. Motivated by these applications, we investigate the constrained preference recommendation (CPR), in which two main types are studied including Threshold-CPR (T-CPR) and Range-CPR (R-CPR). We firstly analyze and define the related problems. Then, we provide solutions with illustration of problem-solving procedure. The results are analogized with other representative techniques. Finally, we provide an extension study on the general CPR problems. Therein, we firstly provide a normalized system for constraint representation. Secondly, we utilize the proposed normalized system to analyze representative literature.
