Content uploaded by Daljeet Singh
Author content
All content in this area was uploaded by Daljeet Singh on Jun 24, 2018
Content may be subject to copyright.
A preview of the PDF is not available
Visualization of Software Metrics for UML Structural and Behavioral Diagrams using Metrics Tool
Figures
Figures - uploaded by Daljeet Singh
Author content
All figure content in this area was uploaded by Daljeet Singh
Content may be subject to copyright.
... To study the architecture of the open-source machine learning software systems, we recovered the package diagram from source code. Apart from these artifacts, we analyzed the architecture using the software engineering metrics, see Figure 1, to describe the characteristics of the architecture [28][29][30][31][32][33][34]. We focus on four main aspects of the architecture, namely, size, coupling, cohesion, and complexity. ...
Machine learning is becoming popular gradually due to the widespread use of AI in different applications ranging from user-specific applications to scientific applications. Machine learning models are usually employed in application programs with the help of open-source machine learning software systems. There is a need to analyze the architecture of such systems. In this work, we chose four different popular open-source machine learning software systems to analyze their architecture. These systems are TensorFlow, Microsoft Cognitive Toolkit, Caffe, Mxnet. We recovered their architecture through a re-engineering approach to building their UML class diagram, Package diagram using the StarUml tool. We analyzed and studied the characteristics of the architecture by using software engineering metrics from the literature. We compared and discussed the open-source machine learning software systems. It is very clear that the architecture quality was not valued heavily in the envision of these systems. In the future , we will investigate more software artifacts and software engineering metrics of these open-source machine learning software systems.
Abstract—In this paper, we keep in mind for increased demand of software quality which has resulted in quality being more of differentiator between products than ever before. The developers need objective and accurate measures for use of improvement of product quality from the previous stages of development of the software. In order to study these metrics for UML structural and behavioral diagrams from different viewpoints, relationships, types of metric values, and their validations. Proper categorization of software design metrics has been formed and with their accurate
The interaction between the classes or within the classes shows the complexity of the design. For one smaller problem, there may be more than one software design but who will be the best; depends on the complexity level of software de-sign. Therefore, coupling and cohesion which shows the interlinking of classes and strength of classes; control the com-plexity of the design. The best software object oriented design is based upon the low coupling and high cohesion level. In the present work, a real case study of Life Insurance policy for handicapped person is demonstrated through the UML Class Diagram; coupling and cohesion levels are measured and results are demonstrated in the form of tables.
This paper proposes a new technique for the measurement of ACOS and ASSOS of the Statechart diagram using Program Slicing. The Statechart diagram contains the dynamic information of the Object-Oriented system. In this research work, State Dependency Graph (STDG) is generated from the Criteria table of the Statechart diagram. Then, the STDG is dynamically sliced, which results easy understandability of statechart diagram for an effective communication among the developers. These slices can be further used to measure accurate ACOS represents cohesion & ASSOS represents coupling, from values of COS and SS respectively. The objective of this new approach is to measure the accurate results of ACOS and ASSOS in contrast to the previous research. General Terms Criteria Table, state dependency graph, dynamic slicing.
Coupling is one of the most important properties that affect the quality of the design and implementation of a software system. In the context of object oriented software development, coupling metrics and their impact on quality attributes have been investigated for a quarter of a century. In this work we review and critically analyze the developments in this domain by considering 26 of the most influential research papers addressing object oriented coupling. Our analysis reveals that a very strong theoretical background has been already developed but unfortunately without a clear impact on the industry practices and software analysis tooling. Even more, recent developments fail to address this problem and seem to even contribute to increasing this gap. We argue that the direction of current research should be shifted towards systematizing and evaluating existing results rather than exploring new applicability domains and defining new metric suites.
Thesis (Ph. D.)--Massachusetts Institute of Technology, Sloan School of Management, 1994. Includes bibliographical references (leaves 243-252).
Measuring Software Design Class Metrics:-A Tool Approach
- Rani Tincy
- Sanyal Manish
- Garg Sushil
Rani Tincy, Sanyal Manish and Garg Sushil, "Measuring
Software Design Class Metrics:-A Tool Approach,"
International journal of engineering research & technology
(IJERT), ISSN: 2278-0181, vol. 1, Issue 7, September 2012.
Determining cohesion and coupling for class diagram through slicing techniques
- Kaur Kumarakhilesh
- Khalsa
KumarAkhilesh and Kaur sunnit khalsa, " Determining cohesion
and coupling for class diagram through slicing techniques",
IJACE, Vol. 4, No..1,pp. 19-24, Jan-June 2012.