KALE-LM: Unleash The Power Of AI For Science Via Knowledge And Logic Enhanced Large Model

2024-09-27

arXiv.org

AI Summary

To address the insufficient reasoning reliability and interpretability of AI in chemical research, this work proposes a large language model (LLM) enhancement paradigm jointly driven by knowledge graph embedding and formal logical constraints, yielding the open-source KALE-LM-Chem(-1.5) model series. Methodologically, it integrates chemical knowledge graphs, first-order logic rule injection, instruction tuning, and chain-of-thought distillation to enable multi-step scientific reasoning and verifiable conclusion generation. Its key contribution lies in being the first to synergistically incorporate structured logical constraints and semantic knowledge embeddings into LLM training—significantly improving accuracy and interpretability on tasks such as reaction prediction and molecular property inference, consistently outperforming general-purpose baselines. The models and code are publicly released, advancing trustworthy AI for Science.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Knowledge Representation and Reasoning: Knowledge Representation Languages
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information

Abstract

Artificial intelligence is gradually demonstrating its immense potential, and increasing attention is being given to how AI can be harnessed to advance scientific research. In this vision paper, we present our perspectives on how AI can better assist scientific inquiry and explore corresponding technical approach. We have proposed and open-sourced two large models of our KALE-LM model series, KALE-LM-Chem(-1.5), which have achieved outstanding performance in tasks related to the field of chemistry. We hope that our work serves as a strong starting point, helping to realize more intelligent AI and promoting the advancement of human science and technology, as well as societal development.

Problem

• Enhancing AI for scientific research using knowledge and logic
• Developing large models for chemistry-related tasks
• Advancing human science and technology with intelligent AI

Innovation

• Knowledge and logic enhanced large model
• Open-sourced KALE-LM-Chem series models
• Outstanding performance in chemistry tasks