XML Keyword Search Based on LLM

As XML data becomes increasingly widespread, traditional query methods like XPath and XQuery struggle with complex and semantic searches. This paper proposes a novel XML keyword search approach using large language models (LLMs) to improve query accuracy and flexibility. By preprocessing XML documents into natural language descriptions, the approach leverages LLMs for better semantic understanding and contextual reasoning. A hybrid method combining inverted indexing and LLM-driven query processing is introduced to balance efficiency and precision. Experimental results demonstrate the effectiveness of this method in handling complex, ambiguous queries, offering a significant improvement over conventional XML query techniques.