A privacy preserving-location_monitoring_system_for_wireless_sensor_networks
Privacy-Preserving Using Data mining Technique in Cloud ...
Transcript of Privacy-Preserving Using Data mining Technique in Cloud ...
Cis-601 Graduate Seminar
Privacy-Preserving Using
Data mining Technique
in Cloud Computing
Submitted by: Rajan Sharma
CSU ID: 2659829
Outline
� Introduction
� Related work
� Preliminaries
� Association Rule Mining with item privacy
� Association Rule Mining with transaction privacy
� Association Rule mining with Database privacy
� Performance Analysis
� Conclusion and future work
Introduction
� Study about the customer behavior
� The discovery of frequent patterns,
association rules, correlation among
huge amount of data is useful to business
intelligence.
This Photo by Unknown Author is licensed under CC BY-SA
Cloud Computing
� The Goal of cloud computing is to
allow users to take benefits from
all these technologies, without
need for deep knowledge.
� The Cloud aims to cut the cost,
and help the user to focus on their
core business instead of being
impeded by IT obstacles.
Data mining-as-a-service in cloud
computing
� In this paradigm, a company (data owner), lacking data storage, computational resources and expertise, stores its data in the cloud and outsources its mining tasks to the cloud service provider (server).
� The association rules mined, k-anonymity, k-support, and k-privacy techniques have been proposed to perturb the data before it is uploaded to the server.
� These techniques are computationally expensive.
� To mine association rules from its data, the user outsources the task to n(≥2) “semi-honest” servers, which co-operate to perform association rule mining on the encrypted data in the cloud and return encrypted association rules to the user.
� we provide three solutions to protecting data privacy during association rule mining.
Data mining-as-a-service in cloud
computing
� Solutions are built on the distributed ElGamal cryptosystem and achieve item
privacy, transaction privacy and database privacy, respectively, as long as at
least one out of the n servers is honest.
� To protect data privacy, the data owner employs the ElGamal cryptosystem to
encrypt all items in a transaction, i.e., the data owner generates its
public/private key pair and then encrypts all items with the public key,
before uploading the transaction onto the cloud.
� When outsourcing the data mining task, the data owner chooses n(n≥2)
different servers in the cloud. It splits its private key into n pieces and
distributes them to the n servers, respectively. The private key is secure as
long as not all the n servers collude
THREE Solutions
Plaintext Equality
Test (PET)
Fake Transactions
Server Counts
Plaintext Equality
Test
� In this n servers can cooperate to
determine the equality of two plaintexts
on the basis of their encryptions without
the need for decryption.
� The basic idea is using PET to identify the
encryptions of the same item in the
encrypted transactions and replace them
with the same encryption.
� Then the Aprori algorithm is applied to the
replacement of the transaction database.
Plaintext Equality
Test
� Problem with this solution that does not hide
the support of each Item set and therefore it
may be vulnerable to the background
knowledge-based attack.
Fake Transaction
� “Wong et el” in 2007 was the first man who
addressed the knowledge based attack and
came up with some idea to prevent it.
� Idea was One to one n item mapping that
transform transactions non deterministically.
� Adding a fake items Into transaction Database.
Fake Transaction
� Example
� I is the set of items in the original database and mis
� a one-to-one mapping from items to integers. The data owner adds a set of fake items F to the dictionary
� J(i.e., |J|=|I|+|F|) and
� maps each item x on to M(x) = {m(x)} ∪ f, where f is a random
� subset of F will be I={a, b, c},m(a) =1, m(b)=2, m(c)=3and F={4,5}, a possible one-to-n item
� mapping M can be defined as M(a) = {1,4,5},M(b) = {2},
� M(c) = {3,5},M(a, b) = M(a)∪M(b) = {1,2,4,5}.