Selected work

Representative Papers

Beyond "AI Language": The case for the idiolectal nature of LLM output

Aug 06, 2026

This study challenges the conventional view that large language model (LLM) outputs constitute a homogeneous “AI language,” arguing instead that individual LLMs exhibit stable, idiolectal linguistic styles akin to human speakers. Drawing on text corpora generated by two distinct LLMs from 2024 and 2026, the research employs computational linguistic descriptors and stylometric principal component analysis (PCA) to empirically assess stylistic variation. Findings reveal significant and consistent differences between models—even under identical prompts—with notable disparities such as a 25-fold variation in contraction usage frequency. These results robustly support the existence of model-specific idiolects, offering novel insights for research in linguistic variation, authorship attribution, and forensic text analysis.

0 citationsRead paper

Exploring the Impact of Generative Artificial Intelligence on Software Development in the IT Sector: Preliminary Findings on Productivity, Efficiency and Job Security

Aug 22, 2025

This study investigates the multifaceted impact of generative artificial intelligence (GenAI) on software development in the IT industry. Employing a mixed-methods design, we first conducted expert interviews to develop a theoretical framework, followed by a large-scale survey (N = XXX) integrating correlational analysis and qualitative insights. Results indicate that 97% of IT professionals currently use GenAI—predominantly ChatGPT—with significant gains in individual productivity. Organizational efficiency exhibits a moderate positive correlation with GenAI adoption intensity (r = .470). However, occupational insecurity concurrently intensifies and shows a strong positive correlation with adoption level (r = .549). This study provides the first empirical evidence of the paradoxical tension between GenAI-driven efficiency gains and eroded job security. Findings offer critical evidence to inform organizational human–AI collaboration strategies and employee adaptive support policies.

0 citationsRead paper

Goat Optimization Algorithm: A Novel Bio-Inspired Metaheuristic for Global Optimization

Mar 04, 2025

To address the imbalance between exploration and exploitation and the tendency to converge prematurely to local optima in metaheuristic algorithms, this paper proposes the Goat Optimization Algorithm (GOA). Inspired by goats’ foraging behavior, parasite-avoidance strategies, and collective coordinated movement, GOA integrates three key mechanisms: adaptive foraging, elite-guided optimal-directional movement, and random jump-based escape. It further introduces a novel solution-filtering mechanism incorporating parasite avoidance to dynamically preserve population diversity and balance exploration versus exploitation. Evaluated on 23 standard benchmark functions, GOA demonstrates statistically significant improvements (Wilcoxon signed-rank test, *p* < 0.05) over PSO, GWO, GA, WOA, and ABC in convergence speed, global search capability, and solution accuracy. Results confirm GOA’s robustness in escaping local optima and its superior performance with statistical significance.

0 citationsRead paper

Can Grammarly and ChatGPT accelerate language change? AI-powered technologies and their impact on the English language: wordiness vs. conciseness

Feb 06, 2025

This study investigates whether AI writing tools—specifically Grammarly and ChatGPT—actively accelerate syntactic simplification in English, using the reduction of the purposive subordinator “in order to” to “to” as a test case. Method: Employing a tripartite methodology—corpus linguistic analysis, NLP-based grammatical error detection, and comparative rhetorical evaluation—we systematically examine editing preferences of these tools on grammatical, natural sentences produced by native speakers. Contribution/Results: Both tools exhibit a statistically significant preference for deleting structurally redundant elements, consistently favoring “to” over “in order to”. This simplification bias operates uniformly across user groups, exerting standardizing pressure toward concision. Crucially, this study provides the first empirical evidence that AI writing assistants do not merely mirror ongoing language change but actively drive syntactic evolution through large-scale, automated editorial intervention. We thus propose “technology-driven linguistic evolution” as a novel mechanism, offering foundational insights for theories of language change and human–AI collaborative writing research.

0 citationsRead paper

Discrete evacuation in graphs with multiple exits

Feb 01, 2025Theoretical Computer Science

This paper addresses the discrete crowd evacuation optimization problem on multi-exit graph-structured environments, aiming to minimize evacuation steps while avoiding inter-agent collisions. Methodologically, it introduces the first formal definition of discrete evacuation equilibrium and an associated optimality criterion tailored to multi-exit scenarios—overcoming limitations of prior single-exit and continuous-flow models. Integrating game-theoretic modeling, graph-based algorithmic optimization, and discrete-event simulation, the authors design a distributed local-decision protocol and establish a rigorous framework for proving global convergence. Experiments on grid graphs, random graphs, and real-world urban road networks demonstrate a 23% reduction in average evacuation time and a 41% decrease in congestion rate, confirming real-time deployability. The core contributions lie in (i) a generalized theoretical framework for discrete evacuation modeling applicable across diverse graph topologies, and (ii) a provably convergent, fully distributed evacuation mechanism.

0 citationsRead paper

Recent publications

Latest Papers

Beyond "AI Language": The case for the idiolectal nature of LLM output

Aug 06, 2026

This study challenges the conventional view that large language model (LLM) outputs constitute a homogeneous “AI language,” arguing instead that individual LLMs exhibit stable, idiolectal linguistic styles akin to human speakers. Drawing on text corpora generated by two distinct LLMs from 2024 and 2026, the research employs computational linguistic descriptors and stylometric principal component analysis (PCA) to empirically assess stylistic variation. Findings reveal significant and consistent differences between models—even under identical prompts—with notable disparities such as a 25-fold variation in contraction usage frequency. These results robustly support the existence of model-specific idiolects, offering novel insights for research in linguistic variation, authorship attribution, and forensic text analysis.

0 citationsRead paper

Exploring the Impact of Generative Artificial Intelligence on Software Development in the IT Sector: Preliminary Findings on Productivity, Efficiency and Job Security

Aug 22, 2025

This study investigates the multifaceted impact of generative artificial intelligence (GenAI) on software development in the IT industry. Employing a mixed-methods design, we first conducted expert interviews to develop a theoretical framework, followed by a large-scale survey (N = XXX) integrating correlational analysis and qualitative insights. Results indicate that 97% of IT professionals currently use GenAI—predominantly ChatGPT—with significant gains in individual productivity. Organizational efficiency exhibits a moderate positive correlation with GenAI adoption intensity (r = .470). However, occupational insecurity concurrently intensifies and shows a strong positive correlation with adoption level (r = .549). This study provides the first empirical evidence of the paradoxical tension between GenAI-driven efficiency gains and eroded job security. Findings offer critical evidence to inform organizational human–AI collaboration strategies and employee adaptive support policies.

0 citationsRead paper

Goat Optimization Algorithm: A Novel Bio-Inspired Metaheuristic for Global Optimization

Mar 04, 2025

To address the imbalance between exploration and exploitation and the tendency to converge prematurely to local optima in metaheuristic algorithms, this paper proposes the Goat Optimization Algorithm (GOA). Inspired by goats’ foraging behavior, parasite-avoidance strategies, and collective coordinated movement, GOA integrates three key mechanisms: adaptive foraging, elite-guided optimal-directional movement, and random jump-based escape. It further introduces a novel solution-filtering mechanism incorporating parasite avoidance to dynamically preserve population diversity and balance exploration versus exploitation. Evaluated on 23 standard benchmark functions, GOA demonstrates statistically significant improvements (Wilcoxon signed-rank test, *p* < 0.05) over PSO, GWO, GA, WOA, and ABC in convergence speed, global search capability, and solution accuracy. Results confirm GOA’s robustness in escaping local optima and its superior performance with statistical significance.

0 citationsRead paper

Can Grammarly and ChatGPT accelerate language change? AI-powered technologies and their impact on the English language: wordiness vs. conciseness

Feb 06, 2025

This study investigates whether AI writing tools—specifically Grammarly and ChatGPT—actively accelerate syntactic simplification in English, using the reduction of the purposive subordinator “in order to” to “to” as a test case. Method: Employing a tripartite methodology—corpus linguistic analysis, NLP-based grammatical error detection, and comparative rhetorical evaluation—we systematically examine editing preferences of these tools on grammatical, natural sentences produced by native speakers. Contribution/Results: Both tools exhibit a statistically significant preference for deleting structurally redundant elements, consistently favoring “to” over “in order to”. This simplification bias operates uniformly across user groups, exerting standardizing pressure toward concision. Crucially, this study provides the first empirical evidence that AI writing assistants do not merely mirror ongoing language change but actively drive syntactic evolution through large-scale, automated editorial intervention. We thus propose “technology-driven linguistic evolution” as a novel mechanism, offering foundational insights for theories of language change and human–AI collaborative writing research.

0 citationsRead paper

Discrete evacuation in graphs with multiple exits

Feb 01, 2025Theoretical Computer Science

This paper addresses the discrete crowd evacuation optimization problem on multi-exit graph-structured environments, aiming to minimize evacuation steps while avoiding inter-agent collisions. Methodologically, it introduces the first formal definition of discrete evacuation equilibrium and an associated optimality criterion tailored to multi-exit scenarios—overcoming limitations of prior single-exit and continuous-flow models. Integrating game-theoretic modeling, graph-based algorithmic optimization, and discrete-event simulation, the authors design a distributed local-decision protocol and establish a rigorous framework for proving global convergence. Experiments on grid graphs, random graphs, and real-world urban road networks demonstrate a 23% reduction in average evacuation time and a 41% decrease in congestion rate, confirming real-time deployability. The core contributions lie in (i) a generalized theoretical framework for discrete evacuation modeling applicable across diverse graph topologies, and (ii) a provably convergent, fully distributed evacuation mechanism.

0 citationsRead paper