Selected work

Representative Papers

Cross-Disciplinary Knowledge Retrieval and Synthesis: A Compound AI Architecture for Scientific Discovery

Nov 23, 2025

The explosive growth of scientific knowledge impedes interdisciplinary knowledge discovery and collaboration. To address this, we propose BioSage—a composite AI architecture integrating large language models (LLMs) with retrieval-augmented generation (RAG) to enable collaborative intelligence across biomedicine, artificial intelligence, data science, and biosafety. Our contributions are threefold: (1) a specialized agent design for cross-disciplinary terminology alignment and traceable reasoning; (2) a modular agent coordination paradigm supporting query planning, response synthesis, cross-modal translation, and multimodal analysis (text, figures, structured data); and (3) a user-centered interactive mechanism. Evaluated on multiple scientific benchmarks, BioSage outperforms baseline LLM and RAG methods by 13–21%. Moreover, on a newly constructed bio-AI cross-modal benchmark, it significantly enhances knowledge acquisition efficiency and research collaboration capability.

0 citationsRead paper

Building Resilient Information Ecosystems: Large LLM-Generated Dataset of Persuasion Attacks

Nov 23, 2025

Generative AI rapidly produces persuasive yet misleading content, undermining governmental and corporate credibility; however, existing institutions lack awareness of and defenses against such persuasive strategies. Method: We introduce the first systematic, multi-model persuasive attack dataset—comprising 134,000 samples across GPT-4, Gemma 2, and Llama 3.1—grounded in the 23 persuasion techniques from SemEval 2023. The dataset features both long and short adversarial texts targeting official government and corporate press releases, with quantitative analysis of model preferences across moral foundations (e.g., Care, Authority, Loyalty). Contribution/Results: Our findings uncover structural biases in large language models’ moral resonance, revealing systematic disparities in ethical alignment across models and domains. This work provides an interpretable theoretical framework and empirically validated insights to support organizational “reputation shielding,” advancing information ecosystem resilience from reactive mitigation toward proactive defense.

0 citationsRead paper

Proactive Defense: Compound AI for Detecting Persuasion Attacks and Measuring Inoculation Effectiveness

Nov 23, 2025

This study addresses the detection and quantification of persuasive attacks targeting human cognition across information environments, with particular emphasis on large language models’ (LLMs) domain-specific vulnerabilities to enhance generative AI safety and human cognitive resilience. Method: We propose BRIES, a composite AI system integrating generative adversarial agents, configurable detectors, content-immunization defenses, and causal inference modules to enable multi-agent collaboration for attack identification, defensive response, and causal attribution. Methodologically, we introduce a persuasion-technique taxonomy grounded in SemEval 2023, a controllable synthetic dataset, and a causal evaluation framework to uncover LLM-specific disparities in rhetorical comprehension. Contribution/Results: Experiments demonstrate GPT-4’s superior performance in detecting complex persuasive techniques; open-source LLMs exhibit significant limitations in fine-grained rhetorical recognition; and temperature settings and prompt engineering critically modulate detection robustness. Code and data are publicly released.

0 citationsRead paper

LLM Enhancement with Domain Expert Mental Model to Reduce LLM Hallucination with Causal Prompt Engineering

Sep 13, 2025

Large language models (LLMs) suffer from hallucination and struggle to integrate complex domain-expert mental models—particularly under training data scarcity—thereby compromising decision-support reliability. Method: We propose a causal prompt optimization framework grounded in Expert Mental Models (EMM). First, we introduce a novel four-step EMM algorithm that pioneers the use of monotonic Boolean functions and multi-valued logic for computationally tractable and scalable mental-structure modeling. Second, we integrate retrieval-augmented generation, causal prompt engineering, human-in-the-loop learning, and factor-wise hierarchical modeling to realize a generalized, EMM-driven personalized prompt generation mechanism. Results: Evaluated on proposal-response decision-making tasks, our framework significantly reduces hallucination rates while improving reasoning consistency and decision-support accuracy—demonstrating both empirical effectiveness and practical applicability in expert-augmented AI systems.

0 citationsRead paper

Recent publications

Latest Papers

Cross-Disciplinary Knowledge Retrieval and Synthesis: A Compound AI Architecture for Scientific Discovery

Nov 23, 2025

The explosive growth of scientific knowledge impedes interdisciplinary knowledge discovery and collaboration. To address this, we propose BioSage—a composite AI architecture integrating large language models (LLMs) with retrieval-augmented generation (RAG) to enable collaborative intelligence across biomedicine, artificial intelligence, data science, and biosafety. Our contributions are threefold: (1) a specialized agent design for cross-disciplinary terminology alignment and traceable reasoning; (2) a modular agent coordination paradigm supporting query planning, response synthesis, cross-modal translation, and multimodal analysis (text, figures, structured data); and (3) a user-centered interactive mechanism. Evaluated on multiple scientific benchmarks, BioSage outperforms baseline LLM and RAG methods by 13–21%. Moreover, on a newly constructed bio-AI cross-modal benchmark, it significantly enhances knowledge acquisition efficiency and research collaboration capability.

0 citationsRead paper

Building Resilient Information Ecosystems: Large LLM-Generated Dataset of Persuasion Attacks

Nov 23, 2025

Generative AI rapidly produces persuasive yet misleading content, undermining governmental and corporate credibility; however, existing institutions lack awareness of and defenses against such persuasive strategies. Method: We introduce the first systematic, multi-model persuasive attack dataset—comprising 134,000 samples across GPT-4, Gemma 2, and Llama 3.1—grounded in the 23 persuasion techniques from SemEval 2023. The dataset features both long and short adversarial texts targeting official government and corporate press releases, with quantitative analysis of model preferences across moral foundations (e.g., Care, Authority, Loyalty). Contribution/Results: Our findings uncover structural biases in large language models’ moral resonance, revealing systematic disparities in ethical alignment across models and domains. This work provides an interpretable theoretical framework and empirically validated insights to support organizational “reputation shielding,” advancing information ecosystem resilience from reactive mitigation toward proactive defense.

0 citationsRead paper

Proactive Defense: Compound AI for Detecting Persuasion Attacks and Measuring Inoculation Effectiveness

Nov 23, 2025

This study addresses the detection and quantification of persuasive attacks targeting human cognition across information environments, with particular emphasis on large language models’ (LLMs) domain-specific vulnerabilities to enhance generative AI safety and human cognitive resilience. Method: We propose BRIES, a composite AI system integrating generative adversarial agents, configurable detectors, content-immunization defenses, and causal inference modules to enable multi-agent collaboration for attack identification, defensive response, and causal attribution. Methodologically, we introduce a persuasion-technique taxonomy grounded in SemEval 2023, a controllable synthetic dataset, and a causal evaluation framework to uncover LLM-specific disparities in rhetorical comprehension. Contribution/Results: Experiments demonstrate GPT-4’s superior performance in detecting complex persuasive techniques; open-source LLMs exhibit significant limitations in fine-grained rhetorical recognition; and temperature settings and prompt engineering critically modulate detection robustness. Code and data are publicly released.

0 citationsRead paper

LLM Enhancement with Domain Expert Mental Model to Reduce LLM Hallucination with Causal Prompt Engineering

Sep 13, 2025

Large language models (LLMs) suffer from hallucination and struggle to integrate complex domain-expert mental models—particularly under training data scarcity—thereby compromising decision-support reliability. Method: We propose a causal prompt optimization framework grounded in Expert Mental Models (EMM). First, we introduce a novel four-step EMM algorithm that pioneers the use of monotonic Boolean functions and multi-valued logic for computationally tractable and scalable mental-structure modeling. Second, we integrate retrieval-augmented generation, causal prompt engineering, human-in-the-loop learning, and factor-wise hierarchical modeling to realize a generalized, EMM-driven personalized prompt generation mechanism. Results: Evaluated on proposal-response decision-making tasks, our framework significantly reduces hallucination rates while improving reasoning consistency and decision-support accuracy—demonstrating both empirical effectiveness and practical applicability in expert-augmented AI systems.

0 citationsRead paper