OR-Star: An Object-Relational Modeling for Heterogeneous Data Warehouse Wen-Yang Lin 1 and Chin-Ang Wu 2 Department of Information Management I-Shou University 1 Hsueh-Cheng Rd, Ta-Hsu, Kaohsiung County, TAIWAN 1 wylin@isu.edu.tw; 2 cwu@csu.edu.tw Abstract Data warehousing is an innovative database concept dedicated to solving the demands of decision support systems. For many years, enterprises have accumulated a lot of data and now realize the importance of using these data for supporting their decision-making. As markets become turbulent, the old way of warehousing data only from the internal operational systems become less relevant in supporting strategic decision. The managers have to acquire and obtain more external data to survive from the competition. This heralds the need of a more powerful data warehousing system that can integrate internal operational data as well as external Internet data, represent and store structured, semi-structured or unstructured data, and provide various data mining tools, to support the strategic decision. We call this Heterogeneous Data Warehouse (Data Warehouse for Heterogeneous Data Sources). In this paper we propose a data model for the heterogeneous data warehouse. This model resembles to the star schema to inherit its easy understanding and multidimensional aspects, and incorporates features of object- relational database to meet the requirement of integrating heterogeneous types of data to support decision making in the Internet era. Keywords. Data warehouse, heterogeneous data, data modeling, object relational database, star schema 1. Introduction The major data sources of a traditional data warehouse are usually from several operational systems inside enterprises. Mostly data are structured in certain format and stored in relational databases or files. Star schema based in relational data model fits well for most cases in managing this kind of data warehouses. Nevertheless, according to an investigation in U.S.A. at 1999 conducted by PWC [8], CEOs of over 400 fast growing companies consider the Internet as the major source of obtaining business intelligence, 82% of high tech companies acquire information of their opponents through Internet, 68% of the surveyed companies use Internet to gain competitive advantages and 27% have direct Internet on line transactions. This survey reveals that traditional data warehouses, which depend primarily on enterprise internal data, cannot provide adequate decision support information for enterprise managers. In the Internet era, business grow more complex and competitive, to face the fiercer competition the company managers must procure more timely information in order to survive in the changeful environment. Therefore, a data warehouse system that can genuinely satisfy the enterprise decision requirements should have the following characteristics: (1) integrate the internal data of an enterprise and external data from the Internet; (2) store and manage structured, semi-structured and even non-structured data; and (3) combine multiple data mining tools to provide decision analyst with suitable and timely information. We refer to this kind of data warehouse as Heterogeneous Data Warehouse (HDW). In this paper, we propose a data model for heterogeneous data warehouse. This model resembles the star schema, and incorporates the technology of object-relational database. In this way, our model not only inherits the advantages of star schema, such as easy understanding and multidimensional modeling, but also meets the requirement of integrating heterogeneous types of data, structured, semi-structured or non- structured, for supporting decision making in the Internet era. The rest of this paper is organized as follows. A review of background material is given in Section 2. In Section 3, we describe the proposed object-relational data model for integrating heterogeneous data. An example is also provided there to illustrate the proposed model. Finally, conclusions are stated in Section 4.