Chintakindi Srinivas

A modified Gaussian similarity measure for clustering software components and documents

An accountable set of dynamic changes are happening on day to day basis in the software industry. So the change is the inevitable and heart of the software industry. Although many software processes and models, tools, standards exist and practices are set, still the industry is facing huge challenge in the design, build and reuse of software components, thereby facing an issue in delivering an effective software product of high quality within a short time meeting customer expectations. Eventually, there is a critical need to throw a light in the direction of understanding related software components and methods to identify similar components so that the components of similar nature may be clustered as a single group. In this paper we propose a novel similarity measure by modifying the Gaussian function. The similarity measure designed is used to cluster the text documents and may be extended to cluster software components and program codes. The similarity measure is efficient as it covers the two sides of the term-axes.

Clustering and Classification of Software Component for Efficient Component Retrieval and Building Component Reuse Libraries

Abstract A Software Repository is a collection of library files and function codes. Programmers and Engineers design develop and build software libraries in a continuous process. Selecting suitable function code from one among many in the repository is quite challenging and cumbersome as we need to analyze semantic issues in function codes or components. Clustering and Mining Software Components for efficient reuse is the current topic of interest among researchers in Software Reuse Engineering and Information Retrieval. A relatively less research work is contributed in this field and has a good scope in the future. In this paper, the main idea is to cluster the software components and form a subset of libraries from the available repository. These clusters thus help in choosing the required component with high cohesion and low coupling quickly and efficiently. We define a similarity function and use the same for the process of clustering the software components and for estimating the cost of new project. The approach carried out is a feature vector based approach.

Document Clustering Using Hybrid XOR Similarity Function for Efficient Software Component Reuse

In this paper a generalized approach is proposed for clustering a set of given documents or text files or software components for reuse based on the new similarity function called hybrid XOR function defined for the purpose of finding degree of similarity among two document sets or any two software components. We construct a matrix called similarity matrix of order n-1 by n for n document sets or software components by applying hybrid XOR function for each pair of document sets. We define and design the clustering algorithm which has its input as similarity matrix and output as a set of clusters formed dynamically as compared to other clustering algorithms that predefine the count of clusters and documents being fit to one of those clusters or classes finally. The approach carried out uses simple computations.

High performance pattern search algorithm using three sliding windows

In this work we propose a new pattern matching algorithm based on the principle of the text segmentation by slicing the text in to three segments. The idea is to perform preprocessing of pattern strings before beginning to search for the pattern in the text so as to achieve substantial speed up in the search process as against to other existing algorithms which either preprocess text or pattern or does no preprocessing such as Brute Force algorithm. The behavior of the algorithm depends on the occurrence of consecutive characters in the event of pattern failure. In this paper we present an efficient pattern matching algorithm based on preprocessing of the pattern string by considering three consecutive characters of the text that immediately follow the aligned pattern window in an event of mismatch between pattern and text character. The algorithm makes use of three sliding patterns. The experimental results show that the proposed algorithm is superior to other algorithms even when the pattern is in the end of the text.

Clustering Text Data Streams – A Tree based Approach with Ternary Function and Ternary Feature Vector☆

Abstract Data is the primary concern in data mining. Data Stream Mining is gaining a lot of practical significance with the huge online data generated from Sensors, Internet Relay Chats, Twitter, Facebook, Online Bank or ATM Transactions. The primary constraint in finding the frequent patterns in data streams is to perform only one time scan of the data with limited memory and requires less processing time. The concept of dynamically changing data is becoming a key challenge, what we call as data streams. In our present work, the algorithm is based on finding frequent patterns in the data streams using a tree based approach and to continuously cluster the text data streams being generated using a new ternary similarity measure defined.

Clustering Software Components for Component Reuse and Program Restructuring

Clustering Software Components for efficient component retrieval is gaining a significant practical importance in the field of software engineering from academic researchers and software industry. Clustering reduces the search space of components by grouping similar entities together thus ensuring reduced time complexity. Finding software components for efficient software reuse is one of the important problems gaining interest from researchers. In this Paper, we first define a similarity function and then give a generalized approach for clustering software components. A component may be a program module or any software document. The objective of component clustering is to form clusters containing high cohesive and low coupling components. Experiments were conducted with Reuters 21578 dataset by considering 70% of documents for training and 30% as test data.

Clustering Software Project Components for Strategic Decisions and Building Reuse Libraries

A Software Repository is a collection of function codes, library files, software requirement specification documents, software design patterns, architectural specifications to name a few. Software Engineers and Programmers analyse, design, implement, develop and build the software libraries, software projects as a continuous process. Mining Software Components for efficient reuse is the current topic of interest among researchers working in the areas of Software Reuse and Information Retrieval. A comparatively less research is contributed in this direction and has a good scope for research. In this paper, the main idea is to cluster the software projects, software components from the available repository and use these clusters in choosing the suitable software component quickly and efficiently. The software clustering process may also be used to estimate and know the hidden knowledge of software systems. We use the similarity function of our previous work submitted at the ACM ISDOC Conference [12] for the purpose of clustering the software projects and software components. The clusters formed may be used to estimate the hidden knowledge and behavior of software projects. The approach carried out is a feature vector based approach.

A Feature Vector Based Approach for Software Component Clustering and Reuse Using K-means

Software component clustering is an unsupervised learning approach which is used to cluster the software components. These clusters may then be used to study, analyze, understand behavior of the software components. In this paper, we use the k-means clustering algorithm to cluster the software components. The main difference lies in the use of distance measure which is designed to find the similarity between the software components. We use the distance measure [12], to find the pair wise project distance matrix and apply the k-means algorithm on this distance matrix. The main idea is to use more than one distance measure, to explore consensus based technique, so as to cluster software components, instead of using only one measure to cluster the components. This approach may also be applied for software architecture recovery problem by using our distance measure.

Feature Vector Based Component Clustering for Software Reuse

Software reuse is concerned about the possibility of reusability of software components. It is important to think about ways, methods and approaches for extracting knowledge from software components. Knowledge mining from software components has various practical applications that range from software reuse to financial applications. This paper proposes a similarity measure for similarity computation between software components by extending our previous research. The proposed measure is used to perform component clustering. Software component clustering facilitates software reusability and software segmentation.