資料庫系統已經是普遍使用的資料儲存工具。企業資料庫中儲存著大量資料，其中隱藏著許多有用的資訊，有效的分析及妥善的運用，可協助企業制訂具有競爭力的策略。管理資訊系統的操作性資料庫無法有效的支援決策分析。近幾年，資料倉儲技術將不同來源的操作性資料庫整合成資料倉庫，以便利支援決策分析。一些研究提出用包含維度綱要及事實綱要的多維度資料模式塑模資料倉庫，然而固定式的維度綱要，不易調整，缺乏彈性及親和性，無法對沙盤推演式資料分析提供良好的支援。本論文探討強化資料庫管理系統對線上資料分析的能力。我們提出以無綱要式的屬性階層擴充關聯資料模式，並且擴充結構化查詢語言，以便利用儲存在屬性階層的資料，進行高層次的資料查詢與分析。另外，我們提出趨勢聚集函數，便利計算區段時間內，數值屬性的資料值變化趨勢。最後，我們開發一視覺化線上資料查詢及分析雛形系統，驗證方法的可行性以及對沙盤推演式資料分析的操作親和性。

Database systems have become a popular tool for data storage. Enterprise’s databases store massive data in that useful information may hide. Properly analyzing those data may help an enterprise in competitive strategies making. Operational databases of management information systems can not effectively support decision making. In recent years, the emerging technology, data warehousing, provides a means to integrate various operational databases into a data warehouse to facilitate on-line data analysis. Some research proposed using multidimensional data model that includes dimension schemata and fad schemata for data warehouse. However, static dimension schemata lack of flexibility and friend1iness can not well support what-if data analysis. This paper discusses strengthening the capability of on-line data analysis of database management systems. We propose extending relational data model with schemaless attribute hierarchy and also extending conventional structured query language in order to utilize the information stored in attribute hierarchies for high-level query and analysis. Besides, we propose a trend aggregation function for computing the trend of a numerical attribute. Finally, we develop a visual on-line query and analysis prototype to prove the feasibility of the proposed model and to show better support to what-if analysis.