Quality and Recommendation of Multi-Source Data for Assisting Technological Intelligence Applications

Abstract

Due to its costly impact, data quality is becoming an emerging domain of research. Motivated by its stakes and issues, especially in the application domain of Technological Intelligence, we propose a generic methodology for modeling and managing data quality in the context of multiple information sources. Data quality has different categories of quality criteria and their evaluations enable the detection of errors and poor quality data. We introduce the notion of relative data quality when several data describe the same entity in the real world but have contradictory values: homologous data. Our approach differs from the general approach for resolving extensional inconsistencies in integration of heterogeneous systems. We cumulatively store homologous data and their quality metadata and we recommend dynamically data with the best quality and data which are the most appropriate to a particular user. A value recommendation algorithm is proposed and applied to the Technological Intelligence application domain.