Research on Intelligent Question-Answer System for Tender Information Based on Knowledge Graph

Traditional methods of obtaining tender information often rely on manual searches or browsing through large volumes of documents and web pages, resulting in low efficiency and the potential for overlooking important details. Therefore, this paper constructs a knowledge graph for tender information and develops a tender information question-answering system based on the tender information knowledge graph, making it more convenient to access tender information. In addition, for small samples of tender information data, this paper proposes an entity relationship triple extraction method based on transfer learning (TL)+BERT+CRF to extract triples from tender information, and stores the extracted knowledge in a Neo4j graph database, building a domain visualization knowledge graph. With the use of named entity recognition and template matching based on the knowledge graph, a tender information question-answering system is formed. The experimental results demonstrate that the triple extraction method adopted in this study shows significantly improved extraction performance compared to the BERT+CRF and BERT+BiLSTM+CRF models in the self-constructed tender information dataset. Furthermore, the question-answering system can also meet users' needs for tender information queries.