AMAmir Hameed Mir

Amir Hameed Mir

*This is an algorithmically generated author record based on publicly available information.

Latest Publications

  • Open Access

    Quantifying Conceptual Evolution: A Novel Framework for Tracking Semantic Drift in Temporal Document Collections

    by Amir Hameed Mir

    We present a novel framework for quantifying and tracking conceptual evolution in temporal document collections through multi-metric semantic analysis. Our methodology introduces three key innovations: (1) ensemble clustering validation combining silhouette coefficient, Calinski-Harabasz index, and Davies-Bouldin score for optimal semantic prototype discovery, (2) permutation-based statistical testing for establishing significant conceptual continuity across time periods, and (3) multi-dimensional conceptual change quantification through centroid shift analysis, distribution divergence via Wasserstein distance, and semantic space transformation measurement. Applied to sustainability discourse spanning 2018-2023, our framework reveals statistically significant paradigm shifts (p < 0.05) with centroid shift magnitudes ranging from 0.142 to 0.387, demonstrating the transition from Corporate Social Responsibility to ESG integration and finally to regulatory-driven net-zero frameworks. The system achieves 94.7% inter-annotator agreement on prototype classification and identifies semantic prototypes with mean intra-cluster coherence of 0.823. Our contributions include rigorous statistical foundations for semantic evolution analysis, automated prototype discovery with validated clustering, and a comprehensive framework for longitudinal discourse analysis applicable across domains from scientific literature to policy documents.

  • Open Access

    Concentric Number Theory: A Geometric Framework for Prime Number Analysis and Classification

    by Amir Hameed Mir

    This paper introduces Concentric Number Theory (CNT), a novel mathematical framework that provides geometric interpretations of number theoretical concepts through concentric ring systems. We establish rigorous axiomatic foundations for CNT and develop systematic methodologies for prime number analysis, including spatial distribution across rings, geometric complexity measures, and clustering patterns. The framework reveals fundamental mathematical properties including the Halving Principle, Perfect Prime Symmetry, and establishes a Prime Complexity Index that correlates with traditional cryptographic strength measures. Through extensive computational experiments analyzing the first 500 primes, we demonstrate that CNT offers new insights into prime distribution, classification, and geometric organization that complement traditional number theoretical approaches. The methodology formalized in this work enables systematic geometric analysis of prime numbers with applications in cryptography, computational mathematics, and prime number theory. All code and data are openly available under CC BY 4.0 license.

  • Open Access

    The Geometry of Truth: Layer-wise Semantic Dynamics for Hallucination Detection in Large Language Models

    by Amir Hameed Mir

    Large Language Models (LLMs) often produce fluent yet factually incorrect statements-a phenomenon known as hallucination-posing serious risks in high-stakes domains. We present Layer-wise Semantic Dynamics (LSD), a geometric framework for hallucination detection that analyzes the evolution of hidden-state semantics across transformer layers. Unlike prior methods that rely on multiple sampling passes or external verification sources, LSD operates intrinsically within the model's representational space. Using margin-based contrastive learning, LSD aligns hidden activations with ground-truth embeddings derived from a factual encoder, revealing a distinct separation in semantic trajectories: factual responses preserve stable alignment, while hallucinations exhibit pronounced semantic drift across depth. Evaluated on the TruthfulQA and synthetic factual-hallucination datasets, LSD achieves an F1-score of 0.92, AUROC of 0.96, and clustering accuracy of 0.89, outperforming SelfCheckGPT and Semantic Entropy baselines while requiring only a single forward pass. This efficiency yields a 5-20x speedup over sampling-based methods without sacrificing precision or interpretability. LSD offers a scalable, model-agnostic mechanism for real-time hallucination monitoring and provides new insights into the geometry of factual consistency within large language models.