查詢結果分析
來源資料
頁籤選單縮合
題 名 | 資料串流頻繁項目集探勘之隱私保護研究=Privacy-Preserving Frequent-Itemset Mining of Data Streams |
---|---|
作 者 | 趙景明; 吳文群; | 書刊名 | 資訊管理展望 |
卷 期 | 10:2 2008.12[民97.12] |
頁 次 | 頁45-65 |
分類號 | 312.1 |
關鍵詞 | 資料探勘; 隱私保護; 頻繁項目集; 資料串流; 漸進式探勘; Data mining; Privacy preservation; Frequent itemsets; Data streams; Incremental mining; |
語 文 | 中文(Chinese) |
中文摘要 | 相較於傳統的靜態資料庫,資料串流的資料具有資料快速流入、資料量龐大、異動頻繁及使用者需要即時回應等特性。由於此新資料型態的興起,近年來資料串流探勘成為一個極為熱門的研究主題。然而,關於資料串流頻繁項目集探勘之安全性研究,目前尚未有相關的研究提出。本論文提出PFSM (Privacy Frequent Itemsets Stream Mining)方法,使用資料匯總(Data Integration)技術進行項目欄位合併來減少資料量並保護資料安全性,並且利用漸進式頻繁項目集探勘(Incremental Frequent Itemsets Mining)演算法之框架位移概念進行快速的頻繁項目集探勘。實驗結果顯示PFSM方法不但兼具安全性,保有良好的精確度,其效能也比傳統演算法更佳。在資料串流的環境下,探勘的效能與安全性都有顯著的改善。 |
英文摘要 | Compared to traditional static databases, data streams have the following characteristics: (1) Data flows in with fast speed; (2) The amount of data is enormous; (3) Data distribution changes constantly with time; (4) Immediate response is required. Due to the emergence of this new type of data, data stream mining has recently become a very popular research issue. There have been many studies proposing efficient mining algorithms for data streams. How-ever, to the best of our knowledge, there is no research that studies the privacy preservation issue of data stream mining. In this paper, we propose a method for privacy-preserving frequent-itemset mining of data streams. We not only use the “data integration” technique to combine items of the data-base for data security, but also use the “incremental frequent itemset mining” algorithm de-signed by the sliding-window model to proceed fast frequent itemset mining. The experiment result shows that our method can solve the problem of privacy-preserving frequent-itemset mining of data streams. It keeps the characteristics of the fast mining, privacy preservation and efficiency. |
本系統中英文摘要資訊取自各篇刊載內容。