– The purpose of this paper is to present an intelligent data-driven framework which provides an effective group-buying aggregation service and thus offers a new opportunity for personalized services in recommendation and advertisement.
– The work presented in the paper analyzes the aggregated group-buying data and creates a compact view of the data which eliminates the potential redundancy and noise. In doing this, the dependencies are discovered from the data in a reverse engineering way. A noise-tolerant method is appreciated, as noise and exception is inevitable in massive data.
– The paper finds that, through the implementation of the intelligent framework, the aggregator will provide a compact view of the group-buying data to customers. According to the empirical study, a 38 percent average decrease of redundancy and noise in the searching results is achieved through the newly built views and corresponding data.
– The paper presents the innovative process of discovering the dependencies and creating views in a data-driven and noise-tolerant way. The proposed intelligent framework improves the aggregation performance and forms the basis of personalized services.
