頁籤選單縮合
題名 | Detecting Rule Behavior Anomalies in Active Databases=主動式資料庫中異常規則行為的偵測 |
---|---|
作 者 | 張志祥; | 書刊名 | 東南學報 |
卷期 | 28 2005.06[民94.06] |
頁次 | 頁59-68 |
分類號 | 312.49 |
關鍵詞 | 主動式資料庫; 規則行為; 衝突更新; 圖論; Active databases; Rule behavior; Contradictory update; Graph theory; |
語文 | 英文(English) |
中文摘要 | 目前的資料庫管理系統都已經提供規則(rule)的設定及規則的處理;系統提供這樣的功能可以將原本被動的處理方式變為一種主動的處理方式。一般來說一個資料庫系統可以設定大量的規則,然而我們比較困難去預測規則的行為及其執行後所產生的效應。當資料庫系統一被更新,相關的規則在交易(transaction)即將結束前將被觸發。在規則的執行過程中,如果發生"衝突更新(contradictory update)"現象時,交易將被中斷(abort)。如果此現象發生的頻率過高時,將會大大降低系統的效能。然而這種現象大部份可以在規則的設定之前就能偵測出來。本篇論文提出一個以圖形為基礎的方法用來找出可能發生衝突更新的規則,以便能事先預防可能發生影響效能的現象。 |
英文摘要 | Various commercial database systems have supported rule specification and processing to bring the traditional passive systems to active ones. A large number of rules may be implemented in a database system. It is very diffcult to predict the behavior of the rules. To process a user-specified update, the related rules may be triggered at the end of the transaction. In the rule execution process, if a phenomenon named contradictory update occurs, the transaction aborts. If the occurrence frequency of the phenomenon is high, the overall system performance will degrade substantially. However, the phenomenon in most cases can be discovered from the interaction between the events and actions specified in the rules. In this paper, we propose a graph-based approach to detect the rule behavior which may inspire the phenomenon such that the superfluous execution of the rules can be avoided. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。