企業會議常產生大量逐字稿、簡報等非結構化資料，蘊含關鍵知識與決策資訊，若未妥善管理將在會後迅速流失。現有智慧會議工具僅能語音轉文字和摘要生成，缺乏對語意、會議脈絡與知識結構的深入理解，無法動態查詢歷史決策或提供推理回應，且缺少內部資料隱私保護機制。
本研究設計並開發一套以大型語言模型（Large Language Model, LLM）為基的「會議知識管理智慧代理人」，整合語意理解、知識圖譜與檢索增強生成 (Retrieval-Augmented Generation, RAG) 技術，以自動擷取、結構化儲存並動態應用會議知識。系統首先對非結構化逐字稿進行語意解析，擷取發言角色、決策事項等語意單元並轉換為結構化中繼資料（Metadata）以建立跨會議知識圖譜。使用Pydantic框架進行資料驗證，並將通過驗證的語意資料寫入雙重儲存架構：組織內部知識文件轉換為語意向量存入FAISS（Facebook AI Similarity Search）向量資料庫，支援語意檢索；會議紀錄逐字稿轉換為節點與關聯寫入Neo4j知識圖譜，建立組織內部的會議知識網路。使用者透過自然語言查詢時，代理人會觸發文件檢索流程，從向量庫與知識圖譜中擷取相關內容，並利用GPT-4模型綜合產出具有上下文脈絡與知識依據的回應。本研究特別強調語意推理能力與決策建議的可解釋性，提示設計亦結合思維鏈提示（Chain-of-Thought Prompting）以提升模型推理邏輯性。藉此代理人不再僅是被動的摘要工具，而成為具備知識推理能力的智慧協作夥伴。
最後為驗證系統效能，本研究建立虛構企業「智鑫科技」之產銷研協調會議資料作為實驗素材，設計多種查詢任務並採用RAGAS指標進行自動化評估。實驗結果顯示在語意檢索準確率（Precision/Recall）均達 0.85 以上，且回應正確率與語意一致性亦具實用水準，證實其在企業決策支援場景中具有實務應用潛力。

In modern enterprises, meetings generate large volumes of unstructured data including transcripts and presentation slides, that contain critical knowledge and decision-related information. If not properly managed, such information is likely to be lost rapidly after the meeting concludes. Existing intelligent meeting tools are typically limited to speech-to-text conversion and basic summarization, lacking in-depth semantic understanding, contextual modeling, and structural knowledge representation. Moreover, they are unable to support dynamic historical decision queries or inferential responses and often fail to ensure the protection of internal data privacy.
This study proposes an intelligent meeting knowledge management agent based on Large Language Model (LLMs), integrating semantic understanding, knowledge graphs, and Retrieval-Augmented Generation (RAG) techniques. The system semantically parses unstructured meeting transcripts to extract key elements such as speaker roles and decision points, transforms them into structured metadata, and validates the data using the Pydantic framework. Verified metadata is stored via a dual-architecture: FAISS for semantic vector retrieval and Neo4j for structured knowledge graph construction. Upon receiving user queries, the agent retrieves relevant content from both sources and uses GPT-4 with Chain-of-Thought prompting to generate contextual and knowledge-grounded responses. Evaluated using simulated enterprise meetings and RAGAS metrics, the system achieved over 0.85 in semantic precision and recall, demonstrating strong potential for practical application in organizational decision support.

Achiam, J., Adler, S., Agarwal, S., Ahmad, L., Akkaya, I., Aleman, F. L., Almeida, D., Altenschmidt, J., Altman, S., & Anadkat, S. (2023). Gpt-4 technical report. arXiv preprint arXiv:2303.08774.
Alstete, J. W., & Meyer, J. P. (2020). Intelligent agent-assisted organizational memory in knowledge management systems. VINE Journal of Information and Knowledge Management Systems, 50(4), 615-630.
Aukščenytė, E., & Jucevičius, G. (2023). Preconditions for the Development of Organizational Knowledge Ecosystem Inside an Audit Firm. ECKM 2023 24th European Conference on Knowledge Managemen Vol 1,
Brown, T., Mann, B., Ryder, N., Subbiah, M., Kaplan, J. D., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., & Askell, A. (2020). Language models are few-shot learners. Advances in neural information processing systems, 33, 1877-1901.
Chen, T., Wang, X., Yue, T., Bai, X., Le, C. X., & Wang, W. (2023). Enhancing abstractive summarization with extracted knowledge graphs and multi-source transformers. Applied Sciences, 13(13), 7753.
Chen, Y., Li, H., Li, H., Liu, W., Wu, Y., Huang, Q., & Wan, S. (2022). An Overview of Knowledge Graph Reasoning: Key Technologies and Applications. Journal of Sensor and Actuator Networks, 11(4), 78. https://www.mdpi.com/2224-2708/11/4/78
De Silva, L., Meneguzzi, F. R., & Logan, B. (2020). BDI agent architectures: A survey. Proceedings of the 29th International Joint Conference on Artificial Intelligence (IJCAI), 2020, Japão.,
Es, S., James, J., Anke, L. E., & Schockaert, S. (2024). Ragas: Automated evaluation of retrieval augmented generation. Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics: System Demonstrations,
Fan, W., Ding, Y., Ning, L., Wang, S., Li, H., Yin, D., Chua, T.-S., & Li, Q. (2024). A survey on rag meeting llms: Towards retrieval-augmented large language models. Proceedings of the 30th ACM SIGKDD conference on knowledge discovery and data mining,
Google LLC. (2025). Enterprise Knowledge Graph overview. Retrieved July 25 from https://cloud.google.com/enterprise-knowledge-graph/docs/overview
Grundspenkis, J. (2007). Agent based approach for organization and personal knowledge modelling: knowledge management perspective. Journal of Intelligent Manufacturing, 18, 451-457.
Guu, K., Lee, K., Tung, Z., Pasupat, P., & Chang, M. (2020). Retrieval augmented language model pre-training. International conference on machine learning,
Han, S., Wang, M., Zhang, J., Li, D., & Duan, J. (2024). A review of large language models: Fundamental architectures, key technological evolutions, interdisciplinary technologies integration, optimization and compression techniques, applications, and challenges. Electronics, 13(24), 5040.
Haz, A. L., Panduman, Y. Y. F., Funabiki, N., Fajrianti, E. D., & Sukaridhoto, S. (2024). Fully Open-Source Meeting Minutes Generation Tool. Future Internet, 16(11), 429. https://www.mdpi.com/1999-5903/16/11/429
He, J., Rungta, M., Koleczek, D., Sekhon, A., Wang, F. X., & Hasan, S. (2024). Does Prompt Formatting Have Any Impact on LLM Performance? arXiv preprint arXiv:2411.10541.
Hogan, A., Blomqvist, E., Cochez, M., d’Amato, C., Melo, G. D., Gutierrez, C., Kirrane, S., Gayo, J. E. L., Navigli, R., & Neumaier, S. (2021). Knowledge graphs. ACM Computing Surveys (Csur), 54(4), 1-37.
Ibrahim, N., Aboulela, S., Ibrahim, A., & Kashef, R. (2024). A survey on augmenting knowledge graphs (KGs) with large language models (LLMs): models, evaluation metrics, benchmarks, and challenges. Discover Artificial Intelligence, 4(1), 76.
Ke, Y. H., Jin, L., Elangovan, K., Abdullah, H. R., Liu, N., Sia, A. T. H., Soh, C. R., Tung, J. Y. M., Ong, J. C. L., & Kuo, C.-F. (2025). Retrieval augmented generation for 10 large language models and its generalizability in assessing medical fitness. npj Digital Medicine, 8(1), 187.
Kirstein, F., Ruas, T., Kratel, R., & Gipp, B. (2024, November). Tell me what I need to know: Exploring LLM-based (Personalized) Abstractive Multi-Source Meeting Summarization. In F. Dernoncourt, D. Preoţiuc-Pietro, & A. Shimorina, Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing: Industry Track Miami, Florida, US.
Kojima, T., Gu, S. S., Reid, M., Matsuo, Y., & Iwasawa, Y. (2022). Large language models are zero-shot reasoners. Advances in neural information processing systems, 35, 22199-22213.
Leidinger, A., van Rooij, R., & Shutova, E. (2023, December). The language of prompting: What linguistic properties make a prompt successful? In H. Bouamor, J. Pino, & K. Bali, Findings of the Association for Computational Linguistics: EMNLP 2023 Singapore.
Lewis, P., Perez, E., Piktus, A., Petroni, F., Karpukhin, V., Goyal, N., Küttler, H., Lewis, M., Yih, W.-t., & Rocktäschel, T. (2020). Retrieval-augmented generation for knowledge-intensive nlp tasks. Advances in neural information processing systems, 33, 9459-9474.
Lima, J., Menezes, L., Teixeira, M., & Redaelli, E. (2023). Perspectives and Trends in Organizational Knowledge Management. In. https://doi.org/10.56238/sevened2023.006-081
López, F., & Cruz, E. (2015). Literature review about Neo4j graph database as a feasible alternative for replacing RDBMS. Industrial Data, 18, 135. https://doi.org/10.15381/idata.v18i2.12106
Masuyama, Y., Chang, X., Zhang, W., Cornell, S., Wang, Z.-Q., Ono, N., Qian, Y., & Watanabe, S. (2025). An end-to-end integration of speech separation and recognition with self-supervised learning representation. Computer Speech & Language, 101813.
Microsoft Corporation. (2025). Overview of Microsoft Graph. Retrieved July 25 from https://learn.microsoft.com/en-us/graph/overview
Motta, R., Barbosa, C. E., Lyra, A., Oliveira, J., Zimbrão, G., & De Souza, J. M. (2022). Extracting Knowledge from and for Meetings. 2022 12th International Conference on Software Technology and Engineering (ICSTE),
Nonaka, I. (1994). A dynamic theory of organizational knowledge creation. Organization science, 5(1), 14-37.
Nova, N. A., & Gonzalez, R. A. (2016). Coordination problems in knowledge transfer: a case study of inter-organizational projects. International Conference on Knowledge Management and Information Sharing,
Park, J. S., O'Brien, J., Cai, C. J., Morris, M. R., Liang, P., & Bernstein, M. S. (2023). Generative agents: Interactive simulacra of human behavior. Proceedings of the 36th annual acm symposium on user interface software and technology,
Qi, P., Huang, Z., Sun, Y., & Luo, H. (2022). A knowledge graph-based abstractive model integrating semantic and structural information for summarizing Chinese meetings. 2022 IEEE 25th International Conference on Computer Supported Cooperative Work in Design (CSCWD),
Seero, L., Burge, J., Soria, A. M., & Van Der Hoek, A. (2023). Exploring a Research Agenda for Design Knowledge Capture in Meetings. 2023 IEEE/ACM 16th International Conference on Cooperative and Human Aspects of Software Engineering (CHASE),
Sun, S., Yuan, R., Li, W., Cao, Z., & Li, S. (2024). Dialogue acts enhanced extract–abstract framework for meeting summarization. Information Processing & Management, 61(3), 103635. https://doi.org/https://doi.org/10.1016/j.ipm.2023.103635
Wachter, S., Mittelstadt, B., & Russell, C. (2024). Do large language models have a legal duty to tell the truth? Royal Society Open Science, 11(8), 240197.
Wei, J., Wang, X., Schuurmans, D., Bosma, M., Xia, F., Chi, E., Le, Q. V., & Zhou, D. (2022). Chain-of-thought prompting elicits reasoning in large language models. Advances in neural information processing systems, 35, 24824-24837.
Yao, S., Zhao, J., Yu, D., Du, N., Shafran, I., Narasimhan, K., & Cao, Y. (2023). React: Synergizing reasoning and acting in language models. International Conference on Learning Representations (ICLR),
Zhao, R., Zhao, F., Wang, L., Wang, X., & Xu, G. (2024). Kg-cot: Chain-of-thought prompting of large language models over knowledge graphs for knowledge-aware question answering. Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence (IJCAI-24),