Volume 17, Issue 5, October 2026

Lawyer Suggestions and Searching for Lawyers Using AI-Powered Recommendation Systems

Rabiya Sarwar, Ali Raza, Junaid Arshad

Department of Computer Science, University of Engineering and Technology (UET), Lahore, Pakistan
College of Law, University of The Punjab, Lahore, Pakistan

Abstract- There is vast information asymmetry in the legal services marketplace: clients struggle to find a lawyer who combines the right expertise, a strong track record, and an affordable cost, since directories like Avvo, LegalZoom and Martindale-Hubbell offer little more than keyword search and subjective peer ratings. The paper proposes an AI-based lawyer recommendation system which integrates a supervised machine learning model for predicting case outcomes with a multi-factor-based recommendation model for matching lawyers and clients. A privacy-preserving synthetic data set of 1000 lawyers and 5000 historical cases was used to train 3 classifiers – Logistic Regression, Random Forest, Gradient Boosting – to predict case outcomes (Win / Lose / Settlement). Gradient Boosting also performed better when evaluated on test accuracy (80.1%), F1 (0.797) and ROC-AUC (0.903) compared with the three classes random baseline of 33.3%. The composite recommendation score of the lawyer quality, win rate and client rating also yielded good ranking quality (NDCG@5 = 0.74, MRR = 0.73, target NDCG@5 = 0.70, target MRR = 0.70). The models were then deployed in a web application built with Flask and exhibited an average response time of 0.9 seconds, and a throughput of 14 queries per second, meeting the predefined targets of ≤2 seconds and ≥10 queries per second. All five research hypotheses were supported by experimental evidence. The results indicate that AI-driven recommendation algorithms can play a crucial role in enhancing the overall accessibility, efficiency, and objectivity of the matching process between lawyers and clients and underscore the need for addressing algorithmic bias, transparency, and data-privacy concerns before these systems can be effectively deployed in practice.

Keywords- Recommendation Systems, Legal Technology, Machine Learning, Lawyer-Client Matching, Gradient Boosting, Synthetic Data Generation and Evaluation Metrics

Download full paper PDF format (Page: 1-8)

Norm Estimates and Characterizations of Bounded Derivations Associated with Hermitian Operators

Babatunde Damilola, Chinara Adebayo

Department of Basic Sciences, University of Science and Technology, Nigeria

Abstract- We examine bounded derivations implemented or induced by Hermitian operators and establish norm estimates that describe their behavior in relation to the underlying operator structure. Particular attention is given to the relationship between the norm of a derivation and the spectral and algebraic properties of the corresponding Hermitian operator. Several conditions under which these derivations admit useful norm bounds and characterizations are discussed. The results provide further insight into the interaction between bounded derivations and Hermitian operators and contribute to the broader theory of operator algebras. The obtained estimates may also serve as useful tools for studying related problems involving commutators, operator inequalities, and derivation-based structures in functional analysis.

Keywords- Bounded Derivations, Algebraic Properties, Hermitian Operators and Functional Analysis

Download full paper PDF format (Page: 9-16)

Harnessing IoT Data for Intelligent Cybercrime Detection and Prevention Using Machine Learning Techniques

Syed Subtain Raza Kazmi

Department of Data Science, University of Engineering and Technology, Lahore, Pakistan

Abstract- The rapid expansion of the Internet of Things (IoT) has transformed modern digital infrastructure by enabling billions of interconnected devices to exchange information and automate critical processes across healthcare, transportation, manufacturing, agriculture, and smart city environments. Although IoT technology offers significant operational and economic benefits, its widespread adoption has also introduced complex cybersecurity challenges. Many IoT devices operate with limited computational resources, weak authentication mechanisms, outdated firmware, and insufficient encryption, making them attractive targets for cybercriminals. Machine Learning (ML) has emerged as a promising approach for strengthening IoT security by identifying malicious activities through data-driven analysis. Unlike conventional security mechanisms that depend primarily on predefined signatures, ML algorithms can learn normal network behavior and detect both known and previously unseen cyber threats. This capability enables faster and more adaptive responses to attacks in dynamic IoT environments. This paper presents a comprehensive review of IoT security vulnerabilities, common cyber threats, and the growing role of machine learning in cybercrime detection and prevention. It examines major attack categories including Distributed Denial-of-Service (DDoS), botnet infections, ransomware, unauthorized access, and data breaches. Furthermore, the study discusses recent advancements in intelligent intrusion detection systems, anomaly detection, explainable artificial intelligence (XAI), and privacy-preserving learning techniques such as federated learning....

Keywords- Internet of Things, Machine Learning, Cybersecurity, Cybercrime Detection, Intrusion Detection Systems, Artificial Intelligence, IoT Security and DDoS Attacks

Download full paper PDF format (Page: 17-24)

An Efficient IoT-Enabled Dual-Mode Learning System for Blind and Deaf-Blind Students

Nimra Nisar, Rabiya Sarwer

----

Abstract- The primary objective of the reviewed system is to enhance information accessibility for visually impaired and deaf-blind individuals by enabling the real-time conversion of printed textual content into tactile Braille output. By leveraging IoT-enabled communication and WSN-based connectivity, the system ensures low-latency, synchronized interaction between input devices, processing modules, and output interfaces, thereby supporting seamless and autonomous user interaction. This review systematically analyzes the architectural design, methodological workflow, and technological components of the proposed system, with particular emphasis on OCR accuracy, Braille translation efficiency, system responsiveness, and communication reliability. In addition to evaluating the technical implementation, the paper presents a comparative analysis with existing assistive technologies, highlighting how the IoT–OCR–Braille integration addresses limitations commonly associated with conventional text-to-Braille and audio-based systems, such as delayed feedback, limited adaptability, and lack of scalability. The findings indicate that the reviewed approach achieves improved recognition accuracy, significantly reduced processing latency, and enhanced adaptability to dynamic environments compared to earlier frameworks. Moreover, the review underscores the broader implications of integrating IoT infrastructure within assistive technologies, emphasizing sustainability, scalability, and long-term deployment feasibility. The system discussed aligns with universal design principles and contributes to digital inclusion by promoting independent learning, improved access to educational resources, and active participation in digital and physical environments for users with sensory impairments. Overall, this review positions the IoT-enabled OCR-Braille solution as a promising and socially impactful model for next-generation assistive technologies, with strong potential for adoption in inclusive education, rehabilitation, and accessibility-oriented smart environments.

Keywords- IoT, Optical Character Recognition (OCR), Wireless Sensor Networks (WSN), Braille Display, Assistive Technology, Accessibility, Visually Impaired, Deaf-Blind, Real-Time Translation, Inclusive Education, Tactile Communication and Smart Devices

Download full paper PDF format (Page: 25-34)

Abstract-

Keywords-

Download full paper PDF format (Page: -)

Abstract-

Keywords-

Download full paper PDF format (Page: - )