| Summary: | 碩士 === 國防大學理工學院 === 資訊科學碩士班 === 97 === The Executive Yuan cooperated with non-governmental organizations have constructed Information Sharing and Analysis Center, ISAC, since 2005. For this reason, the Ministry of National Defense establishes Military ISAC, M-ISAC, which builds up a sharing framework and real-time platform crossing organizations. We can discover useful information from huge raw datasets by using Data mining technologies, among which association analysis is the most popular and powerful tool.
Organizations can obtain more information from data sharing. However, data sharing would potentially involve threats that privacy or sensitive information of the military is unconsciously divulged. Therefore, how to prevent sensitive information being discovered is an important issue now.
In this study, we first proposed HUHSI algorithm for completely hiding sensitive frequent itemsets discovered by association analysis while keeping high utility of the released database. Then, AHSI algorithm was proposed to improve the utility. Finally, taking advantages of evolutional computation, GAHSI algorithm enhanced AHSI algorithm to approach the optima utility. All our algorithms can completely hide sensitive frequent itemsets and avoid new frequent itemsets generated through slightly modifying items of transaction in a released database. Experimental results show that GAHSI outperforms other published algorithms in terms of the utility of released databases.
|
|---|
