Corporates today are facing challenges with their heterogeneous systems. Typical large corporations, public sector and telco companies have around 50-2000 different enterprise systems, ranging from spreadsheet data sources to data marts to data warehouses. The literature study shows that they typically have up to 75% of replication with a data pollution (error rate) occurrence rate of 25%. Our study shows that the intelligent integration of these disparate data sets and systems will help improve business productivity by approximately 40% particularly in day-to-day operations, service management and business reporting. It will also reduce the time wasted on various activities such as logging in and out of different systems and argument the truth of the data report, such as where is the data coming from. However, the integration of these systems is challenged by the fact that data are:
Most importantly, data integration from heterogeneous systems is an urgent operation for enterprises, as more and more legacy systems are due to retire or no longer supported. Having no centralised enterprise data repository means no ownership of the data, no “single source of truth”, and no control of its greatest asset (the enterprise data and information). In addition, there will no automation, no streamline of business processes and policies, no real-time and just-in time data information, no business intelligence and no organization agility.
However, integration of data from heterogeneous sources for corporate data warehouse requires the integration of heterogeneous data from heterogeneous databases and requires integration of heterogeneous workflow operations, processes, policies, and business rules. Without addressing the semantics integration of these, it is impossible to properly integrate the data and it would be just a corporate data dump, rather than Corporate Data Warehouse. Semantic Integration is challenged by the fuzziness of the enterprise heterogeneous data sets across different databases and information systems, and fuzzy interpretation of the concept of data, processes, policies and rules. In this keynote, I present the advanced fuzzy approximate reasoning and defuzzification approaches to semantics data integration in corporate data warehouses, through addressing concept similarities, contradictions, tautology, concept drifting, fuzzy pattern recognition and the evolution development of fuzzy logic enabled ontology based for the enterprise data warehouse that support alignment with human endeavour, IT strategies and business processes. In this talk, I will give 3 case studies, including corporate data warehouse development in UK Defence, Australian Defence force and a Large corporate data warehouse development in an Australian commercial industry.