A Web-based personalized recommendation system for mobile phone selection:Design, implementation, and evaluation

摘要

Recommendation systems that provide appropriate solutions to users to reduce their decision complexity have become popular in the Internet world. Designing and evaluating such systems remain essential challenges to researchers and practitioners. Toward that end, a critical task is how to obtain user prefer-ences. Mobile phones have become indispensable in everyday life, yet fierce market competition, charac-terized by rapid introductions of different models with novel designs and advanced features, have made consumers' purchase decision making increasingly complex. As a well-established, multiple criteria deci-sion technique, analytic hierarchy processing (AHP) provides an intuitive model of a hierarchical struc-ture capable of supporting complex product comparisons and evaluations by consumers. In this paper, we illustrate the application of an AHP-based mechanism to develop a Web-based recommendation sys-tem and empirically evaluate the prototype by conducting a controlled experiment with 244 mobile phone users, focusing on both content and system satisfaction. Our evaluation includes benchmark sys-tems built on rank-based analysis and an equal weight-based system as comparative baselines. Overall, the results suggest the viability and value of using AHP to construct effective recommendation systems. Subjects appear satisfied with the recommendations by the AHP-based system, though its relatively demanding input requirements may need mitigation and adequate interface designs. This study contrib-utes to research and practice in recommender systems in general and helps develop mobile phone rec-ommendation systems for online stores and consumers in particular.