From Abductive Explanations to Global Logical Rules for Node Classification in SGCs
为解决图神经网络预测解释性问题，提出一种基于最小反绎解释的逻辑框架，用于从简单图卷积网络中提取全局逻辑规则。
0 citationsRead paper
Selected work
Representative Papers
Prompt Engineering Strategies for LLM-based Qualitative Coding of Psychological Safety in Software Engineering Communities: A Controlled Empirical Study
This study investigates the reliability of large language models (LLMs) in replicating human reasoning for qualitative coding of psychological safety in software engineering communities, with a focus on performance disparities and systematic biases across different prompting strategies. Through controlled experiments, the authors evaluate Cohen’s κ agreement and stability of Claude Haiku, DeepSeek-Chat, and Gemini 2.5 Flash under zero-shot and few-shot closed-ended prompting. The work presents the first systematic quantification of few-shot prompting effects on LLM-based qualitative coding, revealing that this strategy significantly improves Claude Haiku’s intercoder agreement (Δκ = +0.034). Claude Haiku and DeepSeek-Chat demonstrate the highest stability (SD ≈ 0.017). All models consistently over-predict “sharing negative feedback” and under-predict “expressing concerns,” offering empirical insights and methodological guidance for LLM-assisted qualitative research.
0 citationsRead paper
What Characterizes a Software Leader? Identifying Leadership Practices from Practitioners Social Media
This study addresses a critical gap in software leadership research by moving beyond formal roles or theoretical models to examine how practitioners genuinely enact leadership in practice. Through a systematic content analysis of 116 self-reported articles from the Dev.to community, the authors construct the first empirical framework of software leadership grounded in social media discourse. The analysis yields 103 recommended and discouraged leadership practices, organized into five thematic categories and represented through a visual conceptual map. Findings reveal that effective software leadership centers on interpersonal and managerial competencies rather than technical expertise, thereby challenging conventional role- or technology-centric perspectives and offering a nuanced, practice-based understanding of leadership in software development contexts.
0 citationsRead paper
Concisely Explaining the Doubt: Minimum-Size Abductive Explanations for Linear Models with a Reject Option
This study addresses the problem of generating minimal and faithful abductive explanations in linear models with a reject option. The authors propose efficient algorithms tailored to both acceptance and rejection decisions: a log-linear time algorithm for accepted instances and, for rejected instances, the first 0–1 integer linear programming formulation to compute minimal explanations. This work is the first to achieve minimal abductive explanations in the rejection setting and unifies existing approaches for acceptance within a single coherent framework. Experimental results demonstrate that the proposed methods significantly outperform non-minimality-guaranteed linear programming baselines while preserving explanation minimality and fidelity, thereby enabling efficient, real-time interpretable decision-making in critical applications.
0 citationsRead paper
Bound Propagation meets Constraint Simplification: Improving Logic-based XAI for Neural Networks
This work addresses the computational inefficiency of logic-based explainable artificial intelligence (XAI) methods in large-scale neural networks, which often suffer from high computational costs. The authors propose a novel approach that integrates bound propagation with constraint simplification, uniquely leveraging the results of bound propagation to guide the constraint simplification process. This integration effectively reduces redundant binary variables and tightens neuron value bounds, thereby significantly accelerating the generation of logic-driven explanations. Empirical evaluations demonstrate that the method achieves up to an 89.26% reduction in explanation time on large neural networks, substantially enhancing the scalability and practical applicability of logic-based XAI techniques.
0 citationsRead paper
Recent publications
Latest Papers
From Abductive Explanations to Global Logical Rules for Node Classification in SGCs
为解决图神经网络预测解释性问题，提出一种基于最小反绎解释的逻辑框架，用于从简单图卷积网络中提取全局逻辑规则。
0 citationsRead paper
Prompt Engineering Strategies for LLM-based Qualitative Coding of Psychological Safety in Software Engineering Communities: A Controlled Empirical Study
This study investigates the reliability of large language models (LLMs) in replicating human reasoning for qualitative coding of psychological safety in software engineering communities, with a focus on performance disparities and systematic biases across different prompting strategies. Through controlled experiments, the authors evaluate Cohen’s κ agreement and stability of Claude Haiku, DeepSeek-Chat, and Gemini 2.5 Flash under zero-shot and few-shot closed-ended prompting. The work presents the first systematic quantification of few-shot prompting effects on LLM-based qualitative coding, revealing that this strategy significantly improves Claude Haiku’s intercoder agreement (Δκ = +0.034). Claude Haiku and DeepSeek-Chat demonstrate the highest stability (SD ≈ 0.017). All models consistently over-predict “sharing negative feedback” and under-predict “expressing concerns,” offering empirical insights and methodological guidance for LLM-assisted qualitative research.
0 citationsRead paper
What Characterizes a Software Leader? Identifying Leadership Practices from Practitioners Social Media
This study addresses a critical gap in software leadership research by moving beyond formal roles or theoretical models to examine how practitioners genuinely enact leadership in practice. Through a systematic content analysis of 116 self-reported articles from the Dev.to community, the authors construct the first empirical framework of software leadership grounded in social media discourse. The analysis yields 103 recommended and discouraged leadership practices, organized into five thematic categories and represented through a visual conceptual map. Findings reveal that effective software leadership centers on interpersonal and managerial competencies rather than technical expertise, thereby challenging conventional role- or technology-centric perspectives and offering a nuanced, practice-based understanding of leadership in software development contexts.
0 citationsRead paper
Concisely Explaining the Doubt: Minimum-Size Abductive Explanations for Linear Models with a Reject Option
This study addresses the problem of generating minimal and faithful abductive explanations in linear models with a reject option. The authors propose efficient algorithms tailored to both acceptance and rejection decisions: a log-linear time algorithm for accepted instances and, for rejected instances, the first 0–1 integer linear programming formulation to compute minimal explanations. This work is the first to achieve minimal abductive explanations in the rejection setting and unifies existing approaches for acceptance within a single coherent framework. Experimental results demonstrate that the proposed methods significantly outperform non-minimality-guaranteed linear programming baselines while preserving explanation minimality and fidelity, thereby enabling efficient, real-time interpretable decision-making in critical applications.
0 citationsRead paper
Bound Propagation meets Constraint Simplification: Improving Logic-based XAI for Neural Networks
This work addresses the computational inefficiency of logic-based explainable artificial intelligence (XAI) methods in large-scale neural networks, which often suffer from high computational costs. The authors propose a novel approach that integrates bound propagation with constraint simplification, uniquely leveraging the results of bound propagation to guide the constraint simplification process. This integration effectively reduces redundant binary variables and tightens neuron value bounds, thereby significantly accelerating the generation of logic-driven explanations. Empirical evaluations demonstrate that the method achieves up to an 89.26% reduction in explanation time on large neural networks, substantially enhancing the scalability and practical applicability of logic-based XAI techniques.
0 citationsRead paper