Towards Graph-Based Neuro-Symbolic Logic Reasoning to Improve AI Applications

Summary form only given. Natural language processing (NLP) is a major area of artificial intelligence research, which in its turn serves as a field of application and interaction of a number of other traditional AI areas. Until recently, the focus in AI applications in NLP was on knowledge representation, logical reasoning, and constraint satisfaction - first applied to semantics and later to the grammar. In the last decade, a dramatic shift in the NLP research has led to the prevalence of very large scale applications of statistical methods, such as machine learning and data mining. Naturally, this also opened the way to the learning and optimization methods that constitute the core of modern AI, most notably genetic algorithms and neural networks. In this paper we give an overview of the current trends in NLP and discuss the possible applications of traditional AI techniques and their combination in this fascinating area.

Special Issue of the journal Artificial Intelligence on “AI & Law”

The penetration of large language models (LLMs) into all walks of life requires finding effective ways to improve the accuracy of their practical application in nursing scenarios. This study investigates the potential of retrieval-augmented generation (RAG) technology and the chain of thought (CoT) reasoning process in addressing the limitations of LLMs in professional knowledge and complex problem reasoning. Leveraging a knowledge base derived from the Chinese National Nursing Licensure Examination question bank, researchers first evaluated the baseline performance of LLMs. Subsequently, the CoT reasoning process was systematically compared with official exam parsing methods to assess the model's ability to interpret nursing-related questions. Experimental results demonstrated that integrating the knowledge base significantly improved LLM accuracy from 84.58% to 93.33%. Furthermore, the CoT reasoning process achieved a 91.33% accuracy rate in parsing question options, highlighting its robust logical reasoning capabilities. These findings underscore that the synergistic integration of RAG and CoT enhances the precision of LLMs in knowledge retrieval and clinical reasoning, offering an innovative technical pathway for developing intelligent nursing education tools. The study not only validates the effectiveness of combining knowledge augmentation with advanced reasoning mechanisms but also provides methodological insights for improving the reliability of AI applications in health care.