Research and Application of Large Language Model-Based Similar-Case Recommendation Technology in Electric Power Enterprises

Currently, the technology of similar case recommendation has been widely applied and researched abroad. While its research and application remain relatively limited in China. Based on our research on optimizing natural language processing (NLP) and knowledge graph (KG) technologies. We have adopted a hybrid retrieval strategy to achieve similar case matching and automated recommendation. By using NLP, we enable keyword tokenization, recognition, and contextual comprehension of judgments. We further use KG technology, integrating tags, keywords, and text structures with the unique characteristics of electric power enterprises' legal documents. To enhance similarity-based correlation matching and optimize recommendation accuracy. Through all of the methods, we can automatically process large volumes of case, accurately recommend and predict case outcomes. This provides support for risk management and decision-making in electric power enterprises.