金融服務之時間序列資料管理-模式、技術與建置

Time Series Data Management in Financial Services - Model, Technique, and Implementation

王為孝(Wei-Hsiao Wang)

指導教授 ： 盧以詮 博士

元智大學/管理學院/資訊研究所/碩士(1998年)

摘要

財務金融的時間序列資料管理，因為資料特性導致的儲存問題與關聯式資料庫的特性與限制，一直以來都不見成效。金融研究人員往往使用特定的資料來從事研究，不但無法保證資料的正確性且費時費力。一旦時間久了，對原先資料的產生過程不復記憶。此外，其他研究者並不知已有人從事相同的研究工作而多費心力，這都顯示知識工作者之間「知識分享」的重要性。且資料的產生過程中，研究者往往難以確定資料本身的存續性。曆法系統的建構，正好給於研究者時間性參考的依據。此外，資料來源的多樣性與資料轉換的問題，都在在困擾著使用者。所以本研究的目的在基於「開放式智慧型計算系統」架構下，發展出一個主從式架構的「時間序列資料管理系統」，來提供研究人員資料的轉換、時間序列資料的建立、管理與查詢的便利與正確性。並利用物件導向分析設計之「統合模式語言」為研究方法，從資料、時間、功能和環境模式四方面來分析設計此系統。利用物件導向的元件獨立、可攜性與再使用性，將時間序列系統發展朝向多平台環境的利用。而最後實際的研究成果，「時間序列資料管理系統」也受到廣大金融研究人員的期望與重視。它更可以被視為將來研究多層式「開放式智慧型計算系統」架構的先趨，提供後續發展的便利。

關鍵字

時間序列 ； 物件導向分析設計 ； UML

並列摘要

While the Internet technology and the data availability are rapidly changing, the competition in the investment business becomes intense and increasing. Although the data for financial applications are simple data, the data typically consist of different time series, of which the interrelationships are complex. We were motivated to develop a “Time Series Management System” (TSMS) to support securities investment researchers for their investment decision-makings. The developed TSMS is a knowledge management system featuring with data conversions, time series modeling capabilities, knowledge sharing, computational intelligence, and performance functionalities. We first build up a calendar sub-subsystem for data conversion purposes. We then use the Unified Modeling Language (UML) for data and time modeling. The resulting TSMS, therefore, is a Client/Server architecture under the framework of the proposed OPen Intelligence Computing System (OPICS), which is a component-based with distributed process capabilities.

並列關鍵字

Time Series ； Object-Oriented Analysis and Design ； UML

參考文獻

[2] W. Dreyer, A.K. Dittrich, and D. Schmidt, “An Object-Oriented Data Model for A Time Series Management System,” Proceedings of the 7th International Working Conference on Scientific and Statistical Database Management (SSDBMS''94), Charlottesville, Virginia, Sep. 1994, pp186-195.

[3] R. Chandra, and A. Segev, and M. Stonebraker, “Implementing Calendars and Temporal Rules in Next-Generation Databases,” Proceedings of the 10th Int. Conf. on Data Engineering, Feb, 1994

[6] Yi-Chuan Lu, Hilary Cheng, Calvin Hsu, and Matt Jung, “Financial Decision Support System: A Distributed and Parallel Approach,” Workshop on Distributed System Technologies and Applications, Taiwan, 1999.

[7] D. Schmidt and R. Marti, “Time Series, A Neglected Issue in Temporal Database Research?” Workshop on Temporal Databases, Zurich, Switzerland, Sep 17-18, 1995.

[9] Rational Software Corporation, Rational Rose Product Family, 1997.