Mining and Sharing Heterogeneous E-Marketing Intelligence - A Universal Metadata-Based Approach

Despite the emergent business needs and technical efforts by both industrial and research communities, e-business market intelligence (EMI) still faces many obstacles hindering its successful mining, deployment and sharing, including lack of scaleable and transferable data source, incapability to integrate heterogeneous EMI data, and lack of highly automatic and universally compatible binding techniques. This paper presents a metadata-based approach to facilitate the generation, transformation, sharing and usage of data and knowledge among multiple heterogeneous EMI applications. A metadata-based framework for EMI development has been developed, which provides a range of solutions for mining the metadata from general data sources, and then integrating and migrating the metadata among various data sources. A mechanism has been developed for the automatic binding of metadata and data into EMI applications. A prototype tool was developed to scale up the proposed approach. Case studies have shown that the approach and the prototype tool are efficient and have the potential to deal with industrial level applications