Overwhelming increase in the amount of information raise a requirement of personalized recommendation system. A vast amount of studies have applied traditional collaborative filtering (CF) techniques for generating recommendations. However, well-known scalability issue imposes a limit on its general application. Association Rule based CF techniques where the association rules are primarily generated on items have also been considered but the approaches turned out inefficient with the rapid growth of item space. We anticipated a promising solution of these issues could be diminution in the cardinality of the large user-item rating matrix. Thus, instead of generating associations among the dynamically growing items, generation of associations among the categories, which is quasi-static in practice, could be a convenient route. Herein, we have proposed an elegant approach to generate recommendations using association rule based CF approach on categories. To evaluate the method, we have experimented with real world mobile application user data from Mobilewalla (Mobilewalla is a venture capital backed company which accumulates data for mobile applications from four major platforms, Apple, Android, Windows, and Blackberry). Two new measures are introduced for calculating accuracy. Findings show that our method scales well on dynamic mobile application domain with legitimate accuracy.
Pervin, Nargis; Datta, Anindya; and Dutta, Kaushik, "Towards Generating Recommendations on Large Dynamically Growing Domains" (2013). PACIS 2013 Proceedings. 29.