Research on Intelligent Question Answering Technology for Indicator Procurement Based on Large Language Model

DOI: https://doi.org/10.62517/jbdc.202601101

Author(s)

Longxue Qiao1, Shengchun Xiao1, Fuxian Dou2, Zhicheng Liu3, Shuai Li3,*

Affiliation(s)

1Department of Bidding Procurement, Medical Supplies Center of PLA General Hospital, Beijing, China 2Western Medical Branch of PLA General Hospital, Beijing, China 3Department of Telemedicine, Medical Supplies Center of PLA General Hospital, Beijing, China *Corresponding Author

Abstract

Bidding Q&A and technical consultation are crucial for ensuring fair competition and enhancing transparency in procurement. However, traditional manual and search-engine-based methods struggle with the growing volume of public bidding information. While large language models (LLMs) offer potential for intelligent question-answering systems, their direct application is limited by high fine-tuning costs, tendency to generate hallucinations, and lack of domain-specific accuracy. To address this, this study proposes a hybrid approach integrating Knowledge Graph (KG) and Retrieval-Augmented Generation (RAG) to optimize generative models for intelligent bidding consultation. First, a domain-specific knowledge base is constructed using a KG and jointly learned with an LLM to enhance its professional knowledge representation. Second, the RAG framework dynamically retrieves relevant information from this knowledge base to ground the LLM's responses, thereby improving inference for complex queries. Tests demonstrate that, compared to traditional manual consultation and search engine retrieval, this proposed scheme significantly improves both the accuracy (p<0.05) and response efficiency (p<0.05) of bidding Q&A and technical consultation. The study provides a valuable reference for developing effective intelligent consultation systems in the bidding domain.

Keywords

Big Language Model; Question Answering System; Bidding; Knowledge Graph; RAG

References

[1] Zhang L , Chen L , Liu W .Research and Application of a Question and Answer System for Bidding Based on Knowledge Graph. 2024 2nd International Conference on Signal Processing and Intelligent Computing (SPIC), 2024:928-932.DOI:10.1109/spic62469.2024.10691468. [2] Zhang M , Yang G , Liu Y ,et al.Knowledge graph accuracy evaluation: an LLM-enhanced embedding approach. International Journal of Data Science and Analytics, 2025, 20(3):3021-3035.DOI:10.1007/s41060-024-00661-3. [3] Kim B , Yang J .Comparative Analysis Study on Automated Dataset Generation Frameworks for RAG System Performance Evaluation. The Journal of Korea Institute of Information, Electronics, and Communication Technology, 2025, 18(2):143-154.DOI:10.17661/jkiiect.2025.18.2.143. [4] HAYOU S,GHOSH N,YU B. LoRA+:efficient low rank adaptation of large models.（2024-02-19）[2024-08-01]. https://arxiv. org/abs/2402. 12354. [5] Zhan L , Huang C .Research on Computer Natural Language Processing Intelligent Question Answering System Based on Knowledge Graph. 2024 International Conference on Machine Intelligence and Digital Applications, 2024:70-74.DOI:10.1145/3662739.3664744. [6] Kai Z , Qibo W , Lining W ,et al.Sentiment Analysis of Chinese Bullet Comments Based on Enhanced Deep Learning Model. 2024 IEEE 5th International Conference on Pattern Recognition and Machine Learning (PRML), 2024:481-489.DOI:10.1109/prml62565.2024.10779914. [7] Wang H , Zhang S .Research on the Application of Improved BERT-DPCNN Model in Chinese News Text Classification. CONCURRENCY-PRACTICE AND EXPERIENCE, 2025, 37(3).DOI:10.1002/cpe.8338. [8] Zeng, P., Yuan, LJ., Intelligent Q&A System Based on Bee Knowledge Graph. Information Technology and Informatization, 2023 (7): 108 - 111. doi: 10.3969/j.issn.1672-9528.2023.07.027 [9] Gao W, Zheng X, Zhao S. 2021. Named entity recognition method of Chinese EMR based on BERT-BiLSTM-CRF. Journal of Physics: Conference Series. doi: 10.1088/1742-6596/1848/1 /012083. [10]Shaoxiong J, Shirui P, Erik C, et al. A survey on knowledge graphs: representation,acquisition, and applications. IEEE transactions on neural networks and learning systems, 2021, 33(2): 494-514. [11]Wang Q , Li C , Liu Y ,et al.An Adaptive Framework Embedded With LLM for Knowledge Graph Construction. IEEE transactions on multimedia, 2025:27.DOI:10.1109/ TMM.2025. 3557717. [12]Gürkan ahin, Varol K , Pak B K .LLM and RAG-Based Question Answering Assistant for Enterprise Knowledge Management. 2024 9th International Conference on Computer Science and Engineering (UBMK), 2024:1-6.DOI:10.1109/ubmk63289.2024.10773564. [13]Hamdhana D .A LOW CODE APPROACH TO Q&A ON CARE RECORDS USING FLOWISE AI WITH LLM INTEGRATION AND RAG METHOD. JIPI (Jurnal Ilmiah Penelitian dan Pembelajaran Informatika), 2024, 9(4):2435-2445.DOI:10.29100/jipi.v9i4.6978. [14]Mukhopadhyay P .Designing Conversational Search for Libraries: Retrieval Augmented Generation through Open Source Large Language Models. DESIDOC Journal of Library & Information Technology, 2025, 45(2).DOI:10.14429/djlit.20206. [15]Liao L , Mo W , Wen Y ,et al.Safety Consultation for Prefabricated Construction: A Localized Retrieval-Augmented Generative Question-Answering System. Journal of Computing in Civil Engineering, 2025, 39(5).DOI:10.1061/JCCEE5.CPENG-6315. [16]LIU Z, KUNDU S,LI A,et al. AFLoRA:adaptive freezing of low rank adaptation in parameter efficient fine-tuning of large models.（2024-03-20）[2024-08-01]. https://arxiv. org/abs/2403. 13269. [17] Xing Y , Tan J , Zhang R ,et al.Robust Anomaly Detection of Multivariate Time Series Data via Adversarial Graph Attention BiGRU. BIG DATA AND COGNITIVE COMPUTING, 2025, 9(5):122.DOI:10.3390/bdcc9050122. [18] Guo X , Wang J , Gao G ,et al.Efficient Agricultural Question Classification With a BERT-Enhanced DPCNN Model. IEEE Access, 2024, 12:109255-109268.DOI:10.1109/access. 2024. 3438848.