Cross-domain heterogeneous data query and verification algorithm based on large language models

Addressing the issues of high computational overhead and poor query accuracy in querying and verifying data across different domains and in various formats, this paper proposes a cross-domain heterogeneous data query and verification algorithm based on large language models. The algorithm leverages the semantic understanding and generation capabilities of large language models to construct a unified framework for data query and verification. It employs adapters to fuse cross-domain heterogeneous data and builds corresponding knowledge graphs. The algorithm integrates knowledge graphs and an attention mechanism into the large language model. This integration yields finer-grained semantic structural information. A query rewriting module is designed based on the attention mechanism to transform users' natural language queries into sub-queries that are compatible with each data source, ultimately outputting crossdomain heterogeneous data query and verification results. Experimental results demonstrate that the algorithm achieves high query and verification accuracy, low response times, and high cross-domain trustworthiness across various crossdomain heterogeneous data scenarios. It significantly enhances the efficiency of cross-system data collaboration, providing robust support for the efficient utilization and trustworthy interaction of cross-domain data.