Browsing by Author "Dr. Touseef Tahir"

Now showing 1 - 7 of 7

Automatic Detection of Ambiguities in Functional and Non-Functional Requirements Using Natural Language Processing
(Library Information Services, COMSATS University Islamabad, Lahore Campus, 2024) Fiza; SP22-RCS-018; Dr. Touseef Tahir; LHR TP 9331
Ambiguous user requirements are often perceived as controversial in software engineering. An entire manual approach for dealing ambiguity is a tiresome and time-taking process. The software requirements are essential inputs to software development processes such as software architecture and design, software coding, and software testing. As a consequence of ambiguous requirements, software development professionals such as software architects, software developers, and software testers might develop wrong interpretations. This will cause project cost overrun, delays in project delivery, and quality of software products. The timely identification and correction of requirements ambiguity can lead to better software systems that meet the product objectives and satisfy the needs of stakeholders and end users. This research aims to experiment with natural language processing techniques and propose a model to best detect ambiguities assimilating requirements attributes i.e., completeness and correctness, from functional and non-functional requirements. A dataset of almost 9529 functional and 2400 non-functional requirements from 315 final-year projects will be used to develop the approach. This study will assist project managers to detect ambiguities in the early phases of the software development life cycle in order to address explicit issues of extra time and cost, as well as the accessibility of stakeholders to refine ambiguities during several sessions.
Automatic Generation of Conceptual Models from Functional Requirements using Natural Language Processing
(Library Information Services, COMSATS University Islamabad, Lahore Campus, 2024) Maham Shoaib; SP22-RCS-022; Dr. Touseef Tahir; LHR TP 9330
Conceptual modeling serves as a bridge between functional requirements and the design of a system. It creates a clear and concise model of software requirements that represent the important features and attributes of the system being modeled. Conceptual modeling in natural language processing (NLP) is an open research topic as it enables the developer and the stakeholder to understand and process natural language text, and is essential for many applications such as text classification, information extraction, machine translation, etc. Most existing studies on conceptual modeling have focused on the generation of conceptual models in the form of use cases, class diagrams, activity diagrams, and many more from user stories the problem with the extraction of conceptual from user stories is that user stories are only supported by the agile development process, the dataset used by their approaches consists of maximum 20 to 30 requirements covering only a few domains, data is the principal requirement for the development of any system. Large-sized datasets covering different domains develop efficient and more accurate systems. This thesis research contributes to this research gap. In this research, we have proposed an approach for the automatic extraction of a conceptual model in the form of a class diagram and product goal model from functional requirements using natural language processing techniques. Our approach uses functional requirements for extraction of conceptual models which are supported by almost all development process and a dataset used by our approach consists of 9529 functional requirements from 315 final-year projects, these projects are related to different SDLC models such as waterfall, hybrid, etc., and are supported by different programming languages such as react, Python, etc., and different domains such as ix online services, medical etc. From these different projects in this study, we have covered 4 [hospital management system, online cricket game, online portal for FYP submission, University management system] projects for the creation of a product goal model and 6[hospital management system, online bookstore, online cricket game, online portal for FYP submission, University management system, and online ride booking application] project for creation of class diagram. and each set of projects contains almost 100 to 150 functional requirements Moreover, our approach provides early verification of the conceptual models produced by the designer, which will save time in the code generation and the debugging of the software. To evaluate our approach, automated-generated models are compared with models generated by human experts[5 students of MS computer science] using three evaluation measures: precision, recall, and F1-score. The highest macro-average score was 98%, 82.7%, 87% for elements of the class diagram, and 94%, 86%, and 86% for elements of the product goal model.
Data-driven Predictions of Software Attributes in Future Products
(Library Information Services, COMSATS University Islamabad, Lahore Campus., 2021-11-20) Haseeb Ahmad khan; SP17-BSE-129; Dr. Touseef Tahir
Productions of software systems are increasing rapidly. Software managers have to make estimation calculation and assign specific time period to the different phases of software development process when a project is proposed. The use of expert judgement and intuition in software estimations are mostly inaccurate due to a lack of systematic estimation process. Due to this the company suffers as well as the customer. Our aim is to make a tool that will help the software development companies in data-driven predictions of software attributes (e.g., defects) while using already collected data during development of previous projects e.g., source lines of code (SLOC), defects, function points. The tool will use machine learning and data mining techniques for data-driven prediction and it will automate standard estimation methods of COCOMO, IFPUG and COSMIC. It will help the managers to perform estimations using different graphical user interfaces in the developed software. The tool will also provide a mechanism to use publicly available datasets (e.g., ISBSG and NASA) to build machine learning models that will also help the managers to predict attributes in similar software projects.
Evaluating Classification of Software Requirements using Machine Learning and Natural Learning Processing Approaches
(Library Information Services, COMSATS University Islamabad, Lahore Campus, 2022) Kinza Tasleem; FA20-RCS-019; LHR TP 7911; Dr. Touseef Tahir
The software development process consists of a series of phases e.g., requirement engineering, design, coding, and testing, and each phase is critical for fulfilling the needs of a software user. Requirement engineering is vital to understand, analyze and document the needs and expectations of the user. The functional requirements define the roadmap for the software development process. Functional requirements have not gained attention. No state of art discussed formatting and classification of functional requirements subclasses such as ubiquitous. Optional, unwanted behavior, event driven, and state-driven, and there were no larger datasets publicly available furthermore no datasets were formatted in standard syntax. So, the current research focuses to classify functional requirements subcategories e.g, ubiquitous requirements, event-driven, unwanted behavior, optional features, and state-driven requirements. This research aims to format the requirements using the EARS (Easy Approach to Requirement Syntax) boilerplate and perform several DL, ML techniques, and NLP experiments on a larger dataset of more than 9000 requirements which were created through processing 315 software requirement specifications documents of BS (CS) and BS (SE) final year projects (FYP) of CUI, Lahore to classify functional requirements subclasses. Using natural language processing (NLP) and machine learning (ML) techniques, this study intended to create a framework for classifying functional needs and their subclasses. All software requirements were altered through a series of procedures like normalization, and feature extractions techniques like TF-IDF. Several Machine Learning and Deep Learning experiments were conducted e.g., Logistic Regression (LR), Bernoulli Naïve Bayes (BNB), Decision Tree (DT), Multinomial Naïve Bayes (MNB), Random Forest CNN, and Long Short-Term Memory algorithms to classify functional requirements subclasses. CNN model got a higher result about 0.93 and LSTM achieved an accuracy of 0.92
Exploring Property Price Prediction via Machine Learning and Deep Learning Approaches: The case of Pakistan
(Library Information Services, COMSATS University Islamabad, Lahore Campus, 2022) Ghulam Fatim; SP21-RCS-014; LHR TP 8062; Dr. Touseef Tahir
In many real-world applications, predicting a property price is more realistic and appealing. Property price prediction helps individuals make better decisions when buying or selling a property. However, a large number of realistic tests must be conducted to determine the best methodologies and algorithms to find the optimal combination of these strategies for a reliable property price prediction model. In this research, we created a dataset of around 50k properties of Pakistan. This dataset has been used to train machine learning and deep learning models. In addition, we have applied feature selection methods to identify which attributes of a property can help to predict the best results in terms of evaluation metrics e.g., MAPE, RMSE, MAE, R2 , and MSE. Furthermore, after applying feature selection approaches such as the Pearson correlation coefficient, Mutual Information, Chi-Square, and the VIF on our dataset, we selected those characteristics that improved the reliability of our model. Then we applied ML and DL algorithms and evaluated them using various performance metrics. Afterward, the results revealed that the Extra Trees Reg. performed well as compared to other models of ML and DL. Moreover, after observing the outperforming results of ML & DL models, this research work finds that the data from Pakistan's real estate e-markets can be used for predicting property rates
GetWork – Freelance Market Place
(Library Information Services, CUI Lahore, 2022) Mubashar Ali; sp19-bse- 040; Dr. Touseef Tahir
Machine learning predicts future trends by utilizing hidden data correlations and patterns. The last decade has seen remarkable progress in a fascinating area of machine learning. This method of converting data inputs into predictions has allowed internet giants such as Amazon, Apple, Facebook, and Google to significantly enhance their products. By using machine learning on the data, we can achieve the prediction at maximum accuracy. For the past few years, Freelancing is becoming more and more popular because if someone doesn’t have a job or work to do for earning then he/she must engage with the freelancing and find work according to his/her skills. There are other different freelance platforms exist, which takes high commission rates on projects but our system come up with less fee deduction of only 5% in future but for now our services is free to gather audience for our website.
STUDENT PERFORMANCE PREDICTION SYSTEM FINAL YEAR PROJECT REPORT
(Library Information Services, CUI Lahore, 2023) Hammad Shahzad; FA18-BSE- 054; Dr. Touseef Tahir
Machine learning uses hidden relationships and patterns of data for predicting future trends. Companies use data mining to turn raw data into meaningful information by discovering underlying patterns of data. By using software to look for patterns in large batches of data, organizations and different universities can learn more about their stakeholders and specifically for the students to develop better study strategies, greater results, and lower failure rates. Over the last decade, improvements in processing power and speed have enabled us to automate traditional, time- consuming data analysis. The more complex the data sets collected; the more insights can be gained. We aim to build a data-driven website that uses data mining and machine learning techniques to predict student’s performances according to their historic data. We will first identify student attributes that can be used to predict their performance through analysing available research work. Later, we will collect the data about students’ attributes through a data collection form. The data will be processed by using state of the art data mining methods and by training regression and classification machine learning models. The best performing machine learning model will be used in the website to predict student performance. The website will also enable its users to recommend resources (e.g., books, tutorials, and research papers) based on their experiences while studying a course. It will also enable students to offer their services for understanding a topic or problem in a course