保全序列式與表格式屬性關聯之隱私保護技術

Privacy preserving anonymization for conserving the relation between series and tabular attributes

游博淵(BO-YUAN YOU)

指導教授 ： 戴志華

國立臺北大學/電機資訊學院/資訊工程學系/碩士(2016年)

摘要

在資料探勘的蓬勃發展之下，在各個領域都有得到受惠，但是由於要取得資料所以產生了資料隱私性的問題，而在近年來有許多關於資料保護的技術，都是針對各種不同的資料型態來保護，沒有考量到多種不同的資料型態，所以在本篇論文裡我們提出了CRebSTa方法，這個方法同時考量到表格式的資料和序列式的資料之間的關係，為了保留更多表格式的資料和序列式關聯性，以達到較少的information loss，進而提升較好的utility。

在我們的實驗章節我們的方法與k-anonymity和(k,p)-anonymity比較我們的方法雖然在攻擊保護的部分不是最好的，但我們也取得在三個方法為居中的表現，但是我們的方法在information loss和utility上面，我們的方法是最好的這也說明序列式與表格式資料是需要被同時考量的。

關鍵字

隱私保護 ； 表格式屬性 ； 序列式屬性

並列摘要

Under data mining to flourish in all areas get benefit, but because to get the information so generated data privacy issues, and in recent years many techniques on data protection, are focus one data types to protect, we are focus to different data types, so in this paper we propose CRebSTa method, which at the same time taking into consideration the relationship between tabular data and sequential data between, in order to retain more sequential multi-table data and the relevance of the format, in order to achieve less information loss, and thus enhance the good utility.

In our experimental section of our approach with k-anonymity and (k, p) -anonymity compare our approach, although in attack to protect do not best, but our approach in the performance of three methods to center, but our approach in information loss and utility above, our approach is best which also shows the sequential and tabular data that need to be considered simultaneously.

並列關鍵字

Privacy Preserving ； Tabular Data ； Series Data

參考文獻

[6] Ashwin Machanavajjhala, Daniel Kifer, Johannes Gehrke, and Muthuramakrishnan

[11] Ajumobi Udechukwu, Ken Barker, and Reda Alhajj. Discovering all frequent trends

dimensional k-anonymity. In 22nd International Conference on Data Engineering

beyond k-anonymity and l-diversity. In 2007 IEEE 23rd International Conference

tation of time series, with implications for streaming algorithms. In Proceedings of