Cyber Academia-Chemical Engineering (CA-ChemE): A Living Digital Town for Self-Directed Research Evolution and Emergent Scientific Discovery

TL;DR

The paper tackles the challenge of autonomous, interdisciplinary discovery in chemical engineering by proposing CA-ChemE, a living digital town built on a multi-agent system with domain-specific knowledge bases and a Collaboration Agent governed by ontology engineering. It introduces retrieval-augmented generation, domain-adaptive fine-tuning, and knowledge graphs to empower expert agents, and demonstrates that knowledge-base enhancements improve decision quality by 10–15% while exposing cross-domain collaboration bottlenecks. The Collaboration Agent addresses semantic gaps between domains, yielding substantial improvements for distant-domain pairs (up to 8.5%), and revealing a diminishing collaborative efficiency caused by knowledge-base gaps. Together, these results present a viable pathway toward autonomous, cross-domain scientific discovery in chemical engineering, mediated by structured knowledge integration and semantic bridging across disciplines.

Abstract

The rapid advancement of artificial intelligence (AI) has demonstrated substantial potential in chemical engineering, yet existing AI systems remain limited in interdisciplinary collaboration and exploration of uncharted problems. To address these issues, we present the Cyber Academia-Chemical Engineering (CA-ChemE) system, a living digital town that enables self-directed research evolution and emergent scientific discovery through multi-agent collaboration. By integrating domain-specific knowledge bases, knowledge enhancement technologies, and collaboration agents, the system successfully constructs an intelligent ecosystem capable of deep professional reasoning and efficient interdisciplinary collaboration. Our findings demonstrate that knowledge base-enabled enhancement mechanisms improved dialogue quality scores by 10-15% on average across all seven expert agents, fundamentally ensuring technical judgments are grounded in verifiable scientific evidence. However, we observed a critical bottleneck in cross-domain collaboration efficiency, prompting the introduction of a Collaboration Agent (CA) equipped with ontology engineering capabilities. CA's intervention achieved 8.5% improvements for distant-domain expert pairs compared to only 0.8% for domain-proximate pairs - a 10.6-fold difference - unveiling the "diminished collaborative efficiency caused by knowledge-base gaps" effect. This study demonstrates how carefully designed multi-agent architectures can provide a viable pathway toward autonomous scientific discovery in chemical engineering.

Figures (6)

• Figure 1: Cyber Academia: The Overall Architecture of the Multi-Agent System, showing how different expert agents collaborate through knowledge sharing and efficient cooperation.
• Figure 2: Molecular Design Expert Profile in Cyber Academia, showing the expert's knowledge base architecture, core competencies in molecular design and optimization, and collaborative connections within the multi-agent system.
• Figure 3: Comparison of Expert Dialogue Quality Between Knowledge Base-Enhanced and Non-Knowledge Base Experts. Left panel shows experts without knowledge base support deviating into abstract discussions and failing to solve the technical problem. Right panel demonstrates how knowledge base-equipped experts retrieve relevant literature to provide evidence-based, actionable solutions.
• Figure 4: Performance comparison between knowledge base-enhanced and non-knowledge base expert agents across multiple evaluation dimensions, demonstrating significant improvements in collaboration efficiency, problem-solving accuracy, and adaptability. (a) Overall dialogue scores comparing seven expert agents with and without knowledge base support; (b) Example workflow demonstrating expert agent collaboration with knowledge database; (c) Radar chart for accuracy and precision dimension; (d) Radar chart for response speed dimension; (e) Radar chart for problem-solving ability dimension.
• Figure 5: Comparative Case of CA Coordination Effects. Left: Communication failure between MDE and PSE without CA coordination. Right: Successful collaboration achieved through CA's conceptual conversion and knowledge integration, validating the diminished collaborative efficiency caused by knowledge-base gaps effect mechanism.