Samina Rajper ¹, Zubair A. Shaikh ¹

Affiliations
1 Department of Computer Science, Shah Abdul Latif University Khairpur, Sindh, PK

Abstract

Objectives: The present study is undertaken to survey the Software development cost estimation techniques. This study will provide guidelines and for researchers and practitioners of software engineering. Methods/Analysis: The study was undertaken by planning, conducting and reporting the literature review (LR) for the years 1991-2016. Findings: The study revealed that several SDCE models have been introduced. The reason for the evolution of software cost estimation models may be the changing nature of software complexity, i.e., one cannot exactly predict the cost for the whole project. Not only conventional empirical and quantitative methods but several data mining and machine learning techniques are also used for improved results. However, it is revealed that from quantitative to empirical all SDCE models can be used alone or hybrid with robust ML or DM techniques to estimate the software development exertion.

Keywords

COCOMO, Data Mining Techniques, Machine Learning Techniques, Software Development Cost Estimation.

Full Text

   

Samina Rajper ¹, Noor A. Shaikh ¹, Zubair A. Shaikh ², Ghulam Ali Mallah ¹

Affiliations
1 Department of Computer Science, Shah Abdul Latif University Khairpur, Sindh, PK
2 FAST, Karachi, Sindh, PK

Abstract

Objectives: Automatic detection of E-learners’ learning styles is an important requirement for personalized e-learning. The present study proposes the detection of students’ learning styles automatically on Learning Management System (LMS). Methods/Analysis: The present study proposes different technique of automatic detection of learning styles on LMS using Data Mining technique Bayesian Network (BN). A large survey data is used to map the class room learning styles to E-learning environment which provide significance to incorporated LS model on E-learning systems. Standard questionnaire called Kolb’s Learning Style Inventory (KLSI) is used to identify the students’ learning styles in a class room environment but the proposed technique can automatically detect the learning styles on LMS. Findings: The BN resulted probability values were used as threshold values to detect the learning styles of students in an experiment in which the students of a public university of Pakistan were participated. The participants’ learning styles were found using the manual method and the proposed method. The experiment provided promising results. Novelty/Improvement: Personalized E-learning systems are used to maximize the learning in terms of providing the learning objects as per the students’ requirements. The BN technique is used to replace the KLSI to detect the learners’ learning styles on LMS automatically.

Keywords

Learning Styles, Student Modelling, Bayesian Networks, E-Learning

Full Text

   

Syed I. Jami ¹, Zubair A. Shaikh ¹

Affiliations
1 Mohammed Ali Jinnah University, PK

Abstract

Objective: This paper focuses on integration of Semantic Web with E-Governance by developing a blog based portal for feedback to implement machine readable G2C in an automated way which is one of the targets of Government 3.0. Methods: We propose to change the representation system of blog based portals for linking data in accordance with a general schema to make it semantically meaningful for machines to perform queries. Citizen Feedback system is employed as a test case to extract raw data and associated metadata in an automated way without parsing any structure. Ontology is developed to help in identification of entities and interpret user comments. Findings: The proposed architecture will make it possible for machines to automatically extract and report the general feedback from comments given by a large community of users. Current Web 2.0 based portals show that for estimation of feedbacks, each blog needs to be evaluated either manually or by employing parsing techniques using Bag of Words approach. To implement our proposed approach, we have developed a social semantic-based networking system through which citizens give comments on the policies/decisions implemented or decided to be implemented by the Government. The system interprets machine-readable comments and generates a summary to report the overall public opinion on a certain government policy, hence reducing the gap between citizens and government - one of the primary objectives of eGovernment as Government 3.0. The use of W3C compliant framework in our implementation will lead to universal adaptability of system. It has been found that the proposed approach resulted in a more accurate and efficient analysis of user comments as compared to manual and NLP/Text Mining techniques within a restricted domain. Application: The current implementation can also be extended to cover other domains with a little modification in the ontology schema, such as education, finance, and healthcare.

Full Text