Touseef Tahir

Touseef Tahir


Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
10 July 2025 in Systems
Verified
Journal Article
Open access
PDF file
TTTouseef Tahir
Touseef Tahir
Hamid JahankhaniHamid Jahankhani
Hamid Jahankhani
KTKinza Tasleem
Kinza Tasleem
Bilal HassanBilal Hassan
Bilal Hassan
Software requirements are primarily classified into functional and non-functional requirements. While research has explored automated multiclass classification of non-functional requirements, functional requirements remain largely unexplored. This study addressed that gap by introducing a comprehensive dataset comprising 9529 functional requirements from 315 diverse projects. The requirements are classified into five categories: ubiquitous, event-driven, state-driven, unwanted behavior, and optional capabilities. Natural Language Processing (NLP), machine learning (ML), and deep learning (DL) techniques are employed to enable automated classification. All software requirements underwent several procedures, including normalization and feature extraction techniques such as TF-IDF. A series of Machine learning (ML) and deep learning (DL) experiments were conducted to classify subcategories of functional requirements. Among the trained models, the convolutional neural network achieved the highest performance, with an accuracy of 93, followed by the long short-term memory network with an accuracy of 92, outperforming traditional decision-tree-based methods. This work offers a foundation for precise requirement classification tools by providing both the dataset and an automated classification approach.

1 January 2025 in IET Software
Verified
Journal Article
Open access
TTTouseef Tahir
Touseef Tahir
Bilal HassanBilal Hassan
Bilal Hassan
HJHamid Jahankhani
Hamid Jahankhani
NZNimra Zia
Nimra Zia
MSMuhammad Sharjeel
Muhammad Sharjeel
Automated nonfunctional requirements (NFRs) classification enhances consistency and traceability by systematically labeling requirements, saving effort, supporting early architectural and testing decisions, improving stakeholder communication, and enabling quality across diverse software domains. While prior work has applied natural language processing (NLP) and machine learning (ML) to NFR classification, existing datasets are often limited in size, domain diversity, and contextual richness. This study presents a novel dataset comprising over 2400 NFRs spanning 269 software projects across 26 software application domains, including nine blockchain projects. The raw requirements are standardized using Rupp’s boilerplate to reduce vagueness and ambiguity, and the classification of NFRs types follows ISO/IEC 25,010 definitions. We employ a range of traditional ML, deep learning (DL), and a transformer‐based model (i.e., BERT‐base) for automated classification of NFRs, evaluating performance across cross‐domain and blockchain‐specific NFRs. Results highlight that domain‐aware adaptation significantly enhances classification accuracy, with traditional ML and DL models showing strong performance on blockchain requirements. This work contributes a publicly available, context‐rich dataset and provides empirical insights into the effectiveness of NLP‐based NFR classification in both general and blockchain‐specific settings.

10 August 2025 in Journal of Software: Evolution and Process
Verified
Journal Article
Open access
FTFiza Talha
Fiza Talha
TTTouseef Tahir
Touseef Tahir
TNTalha Nadeem
Talha Nadeem
Ambiguous user requirements present a challenge in software requirement engineering. A manual approach to handling ambiguity is time‐consuming. Software requirements are essential inputs to software development processes, including architecture and design, implementation, and testing. Requirement ambiguities lead to project cost overruns, delays in project delivery, and poor software product quality. Timely identification and correction of ambiguity can result in better software systems that meet product objectives and satisfy the needs of all stakeholders. This study explores various natural language processing techniques and SpanBERT (a variant of BERT). This research proposes a semiautomated approach for detecting anaphoric, coordination, and missing condition ambiguities in functional requirements. The proposed approach is validated on a new, original dataset containing 425 functional requirements from 16 domains. The ambiguities identified through our approach are compared with those detected manually and by ChatGPT. Our approach outperforms ChatGPT in detecting ambiguities. The proposed approach will aid project managers and requirement engineers in identifying ambiguities in requirement specifications, thereby helping to reduce cost overruns and delays in the software development process caused by requirement ambiguities.

9 December 2024
Verified
Conference Paper
Closed access
TTTouseef Tahir
Touseef Tahir
GFGhulam Fatima
Ghulam Fatima
The price prediction is critical for making better decisions when buying or selling a real estate. Currently, there is a lack of data-driven price prediction studies on Pakistani real estate market, which publicly share their datasets and the sources of data collection. In addition, mobile and webapplications of Pakistani electronic markets are also missing the data-driven real estate price prediction. This study collected data of 50 thousand real estate properties of Lahore between 2020 and 2024. The data collected from oldest and most used mobile and web-application. The regression-based machine learning (ML) and deep learning (DL) models are trained on original values of dataset and transformed values (using log and z-score) and subsets of the dataset. The ensemble ML model (i.e., Extra Trees) trained on a subset of features that are selected using mutual information performed the best with evaluation measures of R2 = 0.96, RMSE = 0.09, MAE = 0.05, and MAPE =0.008. The DL model (i.e., Bi-LSTM) trained on logged transformed complete dataset achieved best results based on evaluation measures i.e. R2 = 0.89, RMSE = 0.14, MAE=0.09, MAPE=0.01. The log transformation of feature values helped to achieve best performances of both ML and DL models.

21 June 2026 in Engineering Reports
Verified
Journal Article
Open access
PDF file
MKMuhammad Kashif
Muhammad Kashif
ZHZUNAIB MAQSOOD HAIDER
ZUNAIB MAQSOOD HAIDER
IMIQRA MUNEER
IQRA MUNEER
DKDileep Kumar
Dileep Kumar
TTTouseef Tahir
Touseef Tahir
JSJawad Shafi
Jawad Shafi
Dyslexia is one of the most common learning disabilities and is difficult to detect in the early stages of learning. This is especially complicated in multicultural education systems. An AI-based framework for the detection of dyslexia using handwritten samples of children between the ages of 6 and 10 years is presented in this paper. Due to the scarcity of multilingual dyslexia datasets, we propose a dataset using balanced English-Urdu handwriting containing 852 samples in total. The dataset consists of 426 samples of dyslexic children and 426 samples of non-dyslexic children. Several deep learning architectures using Convolutional Neural Network (CNN), VGG16, InceptionV3, MobileNet1, MobileNet2, and MobileNet3 were used with three optimizers, Adam, SGD, and RMSProp. Evaluation of the optimization of the deep learning methods was suggested to assess the reliability of the results. It was found that the MobileNet1 optimizer with RMSProp managed to achieve results that were overall the best with a high mean accuracy of 0 . 7871 ± 0 . 0102 $$ 0.7871\pm 0.0102 $$ and resulted in the highest mean F 1 $$ {F}_1 $$ -score of 0.7823. These results were especially notable for MobileNet2 with RMSProp and SGD optimizers and a slight decrease in the mean accuracy of the CNN was observed with the SGD optimizer as baseline. This indicates that lightweight deep learning architectures with apt optimizers are effective in capturing handwriting patterns of learners that have dyslexia in a bilingual English–Urdu context. The results go on to show that for small handwriting datasets, repeated-run statistical evaluation offers a more consistent assessment than single-run accuracy reporting. While the framework shows promise to some extent as an early screening support tool, its practical educational or clinical use is contingent upon its large datasets, participant-wise validation, and real-time deployment evaluation. This study advances multilingual dyslexia research by providing a bilingual handwriting dataset and a statistically solid evaluation of lightweight deep learning models for early dyslexia detection.

1 February 2026
Verified
Conference Paper
Closed access
FTFiza Talha
Fiza Talha
TTTouseef Tahir
Touseef Tahir
TNTalha Nadeem
Talha Nadeem
In software requirements engineering, ambiguous non-functional requirements can cause design inconsistencies and poor software quality. Manual ambiguity detection is time consuming and inefficient, particularly for large-scale requirements documents. Prior studies lack in handling ambiguous non-functional requirements and often analyze simple English statements instead of discussing NFRs. In this study, we emphasize detecting anaphoric, coordination, and missing-condition ambiguity in non-functional requirements. For anaphoric ambiguity detection, we applied SpanBERT due to its contextual capturing abilities. Coordination and missing-condition ambiguities were identified using a rule-based approach, i.e., spaCy's Named Entity Ruler. We conducted experiments on a significant dataset consisting of 2400 non-functional requirements from 16 different domains. Experimental results show that the suggested approach performs better than ChatGPT in terms of accuracy and consistency across the evaluated datasets. The proposed approach can assist requirement engineers and project managers in identifying ambiguities at an early stage of development. Timely detection helps reduce rework, associated costs, schedule risks, and improves overall software quality.