An Intelligent Q and A System for Global Petroliferous Basins Based on Knowledge Base and Knowledge Graph Technologies

Since 1993, Chinese oil companies have been participating in global oil and gas exploration and development. Over the past 30 years, a vast number of documents and data on global petroliferous basins has been accumulated through the purchase of international commercial oil and gas databases, independent acquisition of operational area data, and scientific research. Based on this extensive collection of documents and data, we developed a “Knowledge Base for Global Petroliferous Basins.” This knowledge base includes two global-level reports (covering both conventional and unconventional resources), 16 regional-level reports (addressing conventional and unconventional resources in eight regions), and details on 374 conventional petroliferous basins and 99 unconventional petroliferous basins, totaling about 20 GB of information. It encompasses documents and data related to geological characteristics, oil and gas enrichment patterns, and the resource potential of these basins. To enable thorough exploration and user-friendly application of the knowledge base, we created an “Intelligent Question Answering System for Global Petroliferous Basins.” The key technologies involved include document analysis and information extraction, graph databases and graph search capabilities, and a question-answering system based on knowledge graphs. This includes entity mapping, relationship semantic matching, candidate answer retrieval, answer sorting, documentation retrieval, and deep semantic matching. The intelligent QA system addresses the varied needs of users, providing researchers with an intuitive and efficient platform for knowledge mining, with broad applications. Furthermore, establishing a global knowledge graph of petroliferous basins can enhance large language models (LLMs) in the petroleum industry, aiding in detecting LLM hallucinations, knowledge editing, and knowledge injection into the LLM frameworks.