Privacy preserving C4.5 using Gini index

2018

Classification in data mining is the most pervasive problem in the distributed computing environment. It is really challenging to classify the data residing on different sites without revealing the private data to each other. However, decision tree classifier has become a good solution to reduce this problem. This paper explains how to build a privacy preserving decision tree over a vertically partitioned data. The decision tree is created on the data which is partitioned vertically and is owned by different parties along with concealing data of different parties. The proposed algorithm uses a best and most efficient splitting strategy for attributes and at the same time a semi-honest third party is also used. It helps different parties in calculation for construction of decision tree. Scalar product protocol and secure multi-party protocol are the keys behind the security and privacy of the method used. The experimental results show that the accuracy and precision of the proposed a...

In recent years with the development of network data collection and storage technology, the usage and sharing of large amounts of data has become easy process. Once the data and information is accumulated, its will become the wealth of information. Data mining, otherwise known as knowledge discovery, can extracted meaningful information from the large amounts of data because it supports people's decision-making. However the traditional data mining techniques and algorithms directly operated on the original data set, which will cause the leakage of privacy data, these problems challenge the traditional data mining, so privacy-preserving data mining (PPDM) it has become one of the most newest trends in the privacy, security and data mining research. Existing cryptography is the based work for privacy-preserving data mining and is still too slow to be effective for the large scale. But proposed approach is based on exploit for fact that RDTs can naturally it fit into the parallel and fully distributed architecture.

JOURNAL OF COMPUTING, VOLUME 2, ISSUE 1, JANUARY 2010, ISSN 2151-9617, 2010

Data mining deals with automatic extraction of previously unknown patterns from large amounts of data. Organizations all over the world handle large amounts of data and are dependent on mining gigantic data sets for expansion of their enterprises. These data sets typically contain sensitive individual information, which consequently get exposed to the other parties. Though we cannot deny the benefits of knowledge discovery that comes through data mining, we should also ensure that data privacy is maintained in the event of data mining. Privacy preserving data mining is a specialized activity in which the data privacy is ensured during data mining. Data privacy is as important as the extracted knowledge and efforts that guarantee data privacy during data mining are encouraged. In this paper we propose a strategy that protects the data privacy during decision tree analysis of data mining process. We propose to add specific noise to the numeric attributes after exploring the decision tree of the original data. The obfuscated data then is presented to the second party for decision tree analysis. The decision tree obtained on the original data and the obfuscated data are similar but by using our method the data proper is not revealed to the second party during the mining process and hence the privacy will be preserved.

World Academy of Research in Science and Engineering

Privacy preserving data mining (PPDM) is major issue in the areas of datamining and security. PPDM datamining algorithms are analysis the results of impact data privacy. Objective of PPDM is to preserve confidential information by applying datamining tasks without modifying the original data. Because privacy preservation is important for datamining and machine learning applications, it measures to measures designed to protect private information. Numerous datamining techniques are analysis the result of preserved data. But still it becomes loss of information and reduces the utility of training samples. In this research we introduce a decision tree based privacy preserving approach. In this approach the original dataset or data samples are converted into the unrealized dataset where the original data samples cannot be reconstructed if an unauthorized party were to steal some portion. It covers the application of privacy preserving approach with the ID3 decision tree learning algorithm. The problem in existing decision tree based privacy preservation approach is inadequate storage space and it can be implemented only for discrete-valued attributes. Improve the preservation accuracy of the system by building RIPPER learns a rule for a given class; here the examples of that class are denoted as positive instances, examples from the left behind classes are denoted as negative instances. RIPPER algorithm to maximize the information gain and increase the number of rules to covers the non negative rates as well as approach is compatible with other privacy preserving approaches, such as cryptography, for additional protection. In this research we analysis the performance of preserved confidential information with ID3 and RIPPER based approach. A proposed RIPPER result shows the better performance than the existing system.

C4.5 algorithm is developed by Ross Quinlan which is the extension of ID3 algorithm used for generating a decision trees. Since the tree generated by C4.5 can be used for classification, so it's also referred to as statistical classifier. Even though the Random Decision Tree is used to avoid the information leakage there are some problems and issues related to privacy maintenance. When we try to instantiate more instances for one class it leads to ambiguity at the same time creating new classes more and more will increase the complexity in RDT. These problems can be resolved by using our C4.5 algorithm. We can have any number of nodes in a network, each node can create its own tree or class and each class can initiate many number of instances for a disseminated classification consuming secure amount or threshold homomorphic encryption. The main objective of this paper is to discuss the ideal nature of the C4.5 algorithm and how they support this algorithm to be utilized in various datamining process.

International Journal of Engineering & Technology, 2018

In distributed data mining environment maintaining individual data or patterns is a major issue due to high dimensionality and data size. Distributed Data mining framework can help to find the essential decision making patterns from distributed data. Privacy preserving data mining (PPDM) has emerged as a main research area for data confidentiality and knowledge sharing in between the communicating parties. As the distributed data of the individuals are stored by the third party, it leads to the misuse of distributed information in digital networks. Most of the decision patterns generated using the machine learning models for business organizations, industries and individuals has to be encoded before it is publicly shared or published. As the amount of data collected from different sources are increasing exponentially, the time taken to preserve the patterns using the traditional privacy preserving data mining models also increasing due to high computational attribute selection meas...

Journal of Science and Technology on Information security

Data mining has emerged as an important technology for obtaining knowledge from big data. However, there are growing concerns that the use of this technology is infringing on privacy. This work proposes a decision tree mining solution according to the ID3 algorithm that ensures privacy in the 2-Part Fully Distributed setting. Tóm tắt-Khai phá dữ liệu đã nổi lên như một công nghệ quan trọng để thu thập kiến thức từ lượng dữ liệu khổng lồ. Tuy nhiên, ngày càng có nhiều lo ngại rằng việc sử dụng công nghệ này đang vi phạm quyền riêng tư của cá nhân. Bài báo này đề xuất giải pháp khai phá cây quyết định theo thuật toán ID3 có đảm bảo tính riêng tư trong mô hình phân tán đầy đủ hai bên.

International Journal of Advance Engineering and Research Development, 2015

Data mining is a recent advance technology with great potential to help companies focus on the crucial information in the data they have collected about the behavior of their customers. In this paper we are trying to use Data mining as a strong tool for security purposes. This paper covers collection of information within the data that queries and reports can't effectively reveal. Privacy-preserving data publishing addresses the problem of disclosing sensitive data when mining for useful information. In already existing privacy models, differential privacy provides one of the strongest privacy guarantees. In this paper, we have resolved issues like Experimental determination and optimization of the way of sharing information without leaking personal data. Continuous monitoring was being done and with the help of DTI (Decision tree induction classifier) we have classified the upcoming events and made data sharing more reliable. More over the location based identification is also included to ensure the overall security in the future also by resolving this chronic problem.

Proceeding of The International Conference on …

Nowadays, data mining and machine learning techniques are widely used in electronic applications in different areas such as e-government, e-health, e-business, and so on. One major and very crucial issue in these type of systems, which are normally distributed among two or more parties and are dealing with sensitive data, is preserving the privacy of individual's sensitive information. Each party wants to keep its own raw data private while getting useful knowledge from the whole data owned by all the parties. Privacy-preserving Data Mining is dealing with this problem and many protocols have been introduced for various standard data mining algorithms so far. In this paper, we propose some new protocols for two popular techniques, classification and clustering, when data is horizontally or vertically partitioned among two or more parties. In classification we use Gini Index applying on ID3 algorithm to create decision tree from distributed data. We also propose two protocols for k-means Clustering which is a prominent clustering algorithm in data mining techniques. Some secure two and multi-party building blocks such as secure comparison, secure multi-party addition and multiplication are also proposed to use as sub-protocols in our algorithms.

Lecture Notes in Computer Science, 2009

Recently, privacy issues have becomes important in data mining, especially when data is horizontally or vertically partitioned. For the vertically partitioned case, many data mining problems can be reduced to securely computing the scalar product. Among these problems, we can mention association rule mining over vertically partitioned data. Efficiency of a secure scalar product can be measured by the overhead of communication needed to ensure this security. Several solutions have been proposed for privacy preserving association rule mining in vertically partitioned data. But the main drawback of these solutions is the excessive overhead communication needed for ensuring data privacy. In this paper we propose a new secure scalar product with the aim to reduce the overhead communication.