Informatics Publishing Limited

S. Vijayarani ¹

Affiliations
1 Department of Computer Science, Bharathiar University, Coimbatore, Tamil Nadu, IN

Abstract

The concept of sequence data mining aimed to retrieve the frequent patterns in the sequences of products purchased by the customers through the time ordered transactions. Later, the application of sequence mining was extended to complex applications like telecommunication, network detection, DNA and protein sequence research. The technique of searching among sequence data is very important in many applications. A search technique is a technique for finding an item with specified properties among a collection of items. The searching process in sequence databases plays an important role in many application domains, mainly for information retrieval and data mining. When there are a number of stored objects, it will be too slow to linearly search all the stored items to find those that satisfy the query criteria. Hence various techniques and data structures are required to organise and manage the search process so that objects relevant to the query can be located quickly. In this research work, a new sequence search technique SSPP is proposed for performing sequence search operation in a retail dataset. The performance of SSPP technique is analysed to show the efficiency of the proposed technique when compared to other sequence search methods.

Keywords

Sequence, Sequence Database, Prefix Span, SSP, SSPP.

Full Text

     

S. Vijayarani ¹, P. Sathya ¹

Affiliations
1 Department of Computer Science, Bharathiar University, Coimbatore, IN

Abstract

A data stream is a continuous, real time, ordered sequence of items. It is impossible to control the order in which items arrives. Real time surveillances system, telecommunication system, sensor network, financial applications are some of the examples of the data stream systems. These types of streams produced millions or billions of updates every hour. These data must be processed to extract the information in a meaningful way. As data stored in a database and data warehouse are processed by using some mining algorithm. Data mining is an extraction of interesting pattern or knowledge from huge amount of data. In this paper, we will study how the data mining techniques are used in data streams as well as the clustering problem for data stream applications. To partition the data sets into one or more groups of similar objects is known as clustering.

Keywords

Clustering, Data Mining, Data Streams.

Full Text

     

S. Vijayarani ¹, A. Tamilarasi ²

Affiliations
1 School of Computer Science and Engineering, Bharathiar University, Coimbatore, IN
2 Department of MCA, Kongu Engineering College, Perundurai, Erode, IN

Abstract

With the speedy developments in the hardware technology and the rapid escalation of internet increases the capability to accumulate enormous amount of personal data about consumers and individuals. In various circumstances, these data may be mishandled for a variety of purposes. This huge collection of data can be used for data mining also. The concept of data mining is to extract hidden knowledge from the large database. Applying data mining algorithms to get hidden knowledge which is a sensitive one, then it must be safeguarded from others. To perform data mining tasks in a secured way, privacy becomes very vital. Randomization, Statistical Disclosure Control, Cryptography, K-Anonymity and etc. are some of the privacy techniques to perform the data mining tasks in a privacy preserving way. In this paper, we discuss k-anonymity techniques. The inspiring feature at the back of k-anonymity is that many attributes in the data can often be considered as pseudo-identifiers which can be used in conjunction with public records in order to uniquely identify the records. Here, we have experimented the two k-anonymity techniques such as k-anonymity using clustering and de-clustering. Based on the experimental results, we compare the performance of these techniques using ID3 classifier. The result shows that the de-clustering approach provides stronger privacy protection than clustering approach in many circumstances.

Keywords

Privacy, Anonymity, Clustering, De-Clustering, Decision Tree.

Full Text

     

Sivamathi Chokkalingam ¹, S. Vijayarani ²

Affiliations
1 Department of Computer Science, Bharathiar University, Coimbatore, Tamil Nadu, IN
2 Department of Computer Science, School of Computer Science and Engineering, Bharathiar University, Coimbatore, Tamil Nadu, IN

Abstract

Utility mining is an emerging topic in data mining. The aim of utility mining is to discover the itemsets that have maximum utilities. Here utility refers number of items bought, cost of an item or it can be any other user choice in a transaction database. Frequent itemset mining is starting point of utility mining. In frequent itemset mining most often occurring itemsets in a transaction are retrieved. The discovery of such frequent itemsets can help in many business decision making process. Frequent itemset mining concentrates on the number of occurrence of items in a transaction, but not the value of items. But utility mining considers importance of itemsets like the profit it earns in a transaction, quantity in a transaction. In this paper various utility mining algorithms like MEU (Mining with expected utility), FUM (Fast Utility Mining), Two-Phase, CTU-Mine, UP-Growth (Utility Pattern Growth), and FHM (Faster High Utility itemset Mining) MHUI-BIT (Mining High-Utility Itemsets based on BIT vector), MHUT-TID (Mining High-Utility Itemsets based on TIDlist), and THUI (Temporal High Utility Itemsets) are discussed.

Keywords

Utility Mining, High Utility Itemset Mining, MEU, MHUI-BIT & MHUT-TID, THUI-Mine, FUM, Two-Phase CTU-Mine, UP-Growth, FHM.

Full Text

     

R. Janani ¹, S. Vijayarani ¹

Affiliations
1 Department of Computer Science, School of Computer Science and Engineering, Bharathiar University, Coimbatore - 641046, Tamil Nadu, IN

Abstract

Objectives: To retrieve the information after analyzing the contents of the documents which are stored in the desktop by applying string matching algorithms. Methods/Statistical Analysis: To analyze the content of the documents, the various pattern matching algorithms are used to find all the occurrences of a limited set of patterns within an input text or input document. In order to perform this task, this research work used four existing string matching algorithms; they are Brute Force algorithm, Knuth-Morris-Pratt algorithm (KMP), Boyer Moore algorithm and Rabin Karp algorithm. This work also proposes three new string matching algorithms. They are Enhanced Boyer Moore algorithm, Enhanced Rabin Karp algorithm and Enhanced Knuth-Morris-Pratt algorithm. Findings: For experimentation, this work has used two types of documents, i.e. .txt and .docx. Performance measures used are search time, number of iterations and accuracy. From the experimental results, it is realized that the enhanced KMP algorithm gives better accuracy compared to other string matching algorithms. Application/Improvements: Normally, these algorithms are used in the field of text mining, document classification, content analysis and plagiarism detection. In future, these algorithms have to be enhanced to improve their performance and the various types of documents will be used for experimentation.

Keywords

Brute Force, Boyer Moore, Information Retrieval, Knuth-Morris-Pratt, Pattern Matching, Rabin Karp.

Full Text

   

S. Boopathy , N. Mouna , S. Priyanka , S. Vijayarani

Abstract

The resource allocation problem for general multi hop multicast network flows and derives the optimal solution that minimizes the total energy consumption while guaranteeing a statistical end-to-end delay bound on each network path. Existing approaches use negated signal-to-noise ratio as link weights on the complete network graph, finds the minimum spanning tree using those weights to maximize the sum rate, and performs optimal resource allocation on the flow corresponding to the obtained tree structure and maintains a set of dominant flows that are optimal for a potentially large percentage of channel states under a certain network topology and performs flow selection. We propose network flow based algorithm allocates resources in the I th iteration, until all resources are exhausted and the utility is maximized by minimizing the flow cost representing the negative values of 'data rate'. In contrast to maximum-utility resource allocation, the problem of minimum power subject to rate target that we consider does not admit a single-stage multi-commodity flow formulation. In the proposed NFBA, we maximize the 'potential power saving' on the flow instead of minimizing the cost, in a Ft adaptively. We analyze our proposed scheme in terms of complexity, power and cost.

Full Text

   

Janani Balakumar ¹, S. Vijayarani ¹

Affiliations
1 Bharathiar University, Coimbatore, Tamil Nadu, IN

Abstract

Cluster analysis is an unsupervised learning approach that aims to group the objects into different groups or clusters. So that each cluster can contain similar objects with respect to any predefined condition. Text document clustering is the important technique of text mining in efficiently organizing the large volume of documents into a small number of significant clusters. The main objective of this research work is to cluster the collection of documents into related groups based on the contents of the particular documents. In order to perform this clustering task, this research work makes use of two existing algorithms, namely K-means and Bisecting K-means algorithm, and also this research work proposes a new clustering algorithm namely Enhanced-Bisecting K-means algorithm. From the experimental results it is observed that the proposed algorithm gives the better clustering accuracy than other algorithms.

Keywords

Text Mining, Text Document Clustering, K-Means, Bisecting K-Means, Enhanced Bisecting K-Means.

Full Text

     

C. Sivamathi ¹, S. Vijayarani ²

Affiliations
1 Department of Computer Science, Bharathiar University, Coimbatore - 641046, Tamil Nadu, IN
2 Department of Computer Science, Bharathiar University, Coimbatore - 641046, Tamil Nadu,, IN

Abstract

Objectives: Utility mining gains more attraction in recent years. Utility mining can be defined as mining item’s utility and revealing high utility items. In this work an algorithm EHUFIM (i.e Enhanced High Utility Frequent Itemset Mining), was proposed to reveal high utility – frequent itemsets even with negative profits. Methods/Statistical Analysis: The proposed algorithm uses utility mining methods to retrieve high profitable items. Then it uses support measure to reveal high occurrence items. The algorithm implements filter procedure of HUINIV algorithm to handle negative profit items. Findings: This algorithm discovers itemset that have more frequency and high utility with negative profit. Discovering such items helps in decision making in super markets, cross product marketing etc. The proposed algorithm was executed and performance of the algorithms was calculated. Application/Improvement: Existing utility frequents mining algorithms does not consider negative profit values. But, this proposed algorithm, takes negative utility values into account.

Keywords

High Utility Itemset, Negative Utility Itemset, Transaction Weighted Utility Property, Utility Frequent Itemset Mining, Utility MiningHigh Utility Itemset, Negative Utility Itemset, Transaction Weighted Utility Property, Utility Frequent Itemset Mining, Utility Mining

Full Text

   

S. Vijayarani ¹, C. Sivamathi ², S. Maria Sylviaa ³

Affiliations
1 Department of Computer Science, School of Computer Science and Engineering, Bharathiar University, Coimbatore, IN
2 Department of Computer Science, PSG College of Arts & Science, Coimbatore, IN
3 Department of Computer Science, Nirmala College for Women, Coimbatore, IN

Abstract

Dimensionality Reduction is one of the useful techniques used in number of applications in order to reduce the number of features to improve the productivity and efficiency of the task. Clustering is one of the influential tasks in data mining. Dimensionality reductions are used in data mining, Image processing, Networking, Mobile computing, etc. The elementary intention of this work is to apply dimensionality reduction algorithms and then cluster the datasets to detect outliers. A bio-inspired ACO (Ant Colony optimization) algorithm has been proposed to reduce dimensionality. Also another bio-inspired algorithm FA (Firefly Algorithm) has been proposed to detect outliers. The three distinct medical datasets: thyroid dataset, Oesophagal dataset and Heart disease dataset are used for experimental results.

Keywords

Dimensionality Reduction, Clustering, Outlier Detection, ACO (Ant Colony Optimization) Algorithm, FA (Firefly Algorithm).

Full Text