資料倉儲的資料模式大都是以關連式資料模式為主,但近幾年物件關連式資料模式已開始被應用在多媒體資料庫、華爾街新聞資料庫等系統中,用以處理較複雜的資料及查詢。因此,為了提供資訊輔助決策,有必要將物件觀念導入關連式模式中並且建立相應之資料倉儲。 本研究首先提出一物件關連式資料倉儲(Object Relational Data Warehouse,ORDW)的概念層模式(Conceptual Model),稱為EDF Model(Extended Dimension Fact Model),用以擴張資料倉儲系統的資料處理型態,使能表達繼承、子類別與母類別的概念。同時發展一套演算法,可將擴充式個體關係(Extended Entity Relationship Model)圖轉換成EDF Model,最後藉由Project Inc.以及Hospital二個實際例子的EER Model驗證此半自動轉換演算法;此EDF Model與半自動轉換演算法可作為物件關連式資料倉儲綱要設計的參考依據。 因此本研究不但把資料倉儲帶入新的研究領域,亦為資料倉儲設計之半自動化建立新的里程。
The data models for data warehouses have been mostly relational for decades. Object-relational data modeling, however, has started to gain much attention in areas like multimedia databases, geographic information systems, and so on. As the use of multimedia data and unstructured (or semi-structured) data are becoming more common, it will be necessary to include those complex data in the data warehouses. Therefore, introducing the object concepts to relational data modeling for data warehousing is a new way to build a data warehouse, which can handle more complex data and process more complex queries. The objective of this project is to propose a concept model for object relational data warehouse that allows the data warehouse to handle complex data. Furthermore, we develop an algorithm to automatically transform the extended entity relationship model of the source databases to the proposed concept model to make our approach a viable solution.