| dc.contributor.advisor | 蔡子傑 | zh_TW |
| dc.contributor.advisor | Tsai, Tzu-Chieh | en_US |
| dc.contributor.author (Authors) | 鄞雋衡 | zh_TW |
| dc.contributor.author (Authors) | Yin, Chun-Heng | en_US |
| dc.creator (作者) | 鄞雋衡 | zh_TW |
| dc.creator (作者) | Yin, Chun-Heng | en_US |
| dc.date (日期) | 2025 | en_US |
| dc.date.accessioned | 4-Aug-2025 13:57:33 (UTC+8) | - |
| dc.date.available | 4-Aug-2025 13:57:33 (UTC+8) | - |
| dc.date.issued (上傳時間) | 4-Aug-2025 13:57:33 (UTC+8) | - |
| dc.identifier (Other Identifiers) | G0111753135 | en_US |
| dc.identifier.uri (URI) | https://nccur.lib.nccu.edu.tw/handle/140.119/158474 | - |
| dc.description (描述) | 碩士 | zh_TW |
| dc.description (描述) | 國立政治大學 | zh_TW |
| dc.description (描述) | 資訊科學系 | zh_TW |
| dc.description (描述) | 111753135 | zh_TW |
| dc.description.abstract (摘要) | 因應智慧商圈資訊爆炸與即時推播需求，傳統中心化發布/訂閱系 統在彈性與擴展性上遇到瓶頸。Named Data Networking(NDN)雖具 備去中心化與內容導向優勢，但其結構化命名機制難以支援自然語言 訂閱需求。本研究提出整合大型語言模型(LLM)與 NDN 的混合式 架構，設計「智慧商圈命名約定」(SBD-NC)，並在邊緣運算架構上實 現語意解析，將用戶查詢自動轉換為結構化命名，解決命名自由度與 一致性問題。實驗結果顯示，本系統於語意轉換準確率、效能與擴展 性均優於傳統方法，展現去中心化架構於動態物聯網環境的應用潛力。 | zh_TW |
| dc.description.abstract (摘要) | In response to the information explosion and real-time push demands in smart business districts, traditional centralized publish/subscribe systems encounter bottlenecks in flexibility and scalability. Although Named Data Networking (NDN) offers the advantages of decentralization and content- orientation, its structured naming mechanism struggles to support natural lan- guage subscription demands. This research proposes a hybrid architecture integrating Large Language Models (LLMs) with NDN, designing a ”Smart Business District Naming Convention” (SBD-NC) and implementing seman- tic parsing on an edge computing architecture to automatically convert user queries into structured names, addressing the challenges of naming freedom and consistency. Experimental results demonstrate that our system surpasses traditional methods in semantic conversion accuracy, performance, and scal- ability, showcasing the potential of a decentralized architecture in dynamic IoT environments. | en_US |
| dc.description.tableofcontents | 1 緒論 1
1.1 背景與動機 1
1.2 研究目的 2
2 相關文獻 3
2.1 發布/訂閱系統 3
2.2 命名資料網路 5
2.3 NDN發布/訂閱系統 8
2.4 模糊語義匹配 13
3 方法與設計 15
3.1 架構總覽 15
3.1.1 邊緣運算與NDN整合架構 15
3.1.2 服務流程 17
3.1.3 自然語言命名挑戰導覽 19
3.2 自然語言轉換的命名挑戰 20
3.2.1 命名自由度衝突 20
3.2.2 命名一致性問題 21
3.3 命名方法設計 22
3.3.1 命名結構與分隔符號 22
3.3.2 命名元件 22
3.4 邊緣節點設計 25
3.4.1 模組設計 25
3.4.2 資料流程 28
3.5 語意處理流程 30
3.5.1 第一階段:Preprocessing 30
3.5.2 第二階段:LLM深度理解 32
3.5.3 第三階段:語義標準化與智能修正 33
3.5.4 第四階段:結構化輸出生成 34
3.6 NDN封包封裝 36
4 實驗設置與成果 37
4.1 實驗環境 37
4.2 實驗一:語意轉換準確度 39
4.2.1 實驗設計 39
4.2.2 評估指標 40
4.2.3 實驗結果 41
4.3 實驗二:命名顆粒度對NDNPub/Sub系統的影響 43
4.3.1 實驗設計 43
4.3.2 評估指標 43
4.3.3 實驗結果 44
4.4 實驗三:多場景規模效能比較 48
4.4.1 實驗設計 48
4.4.2 實驗參數 49
4.4.3 評估指標 49
4.4.4 實驗結果 50
5 結論與未來工作 53
5.1 結論 53
5.2 未來工作 54
參考文獻 55 | zh_TW |
| dc.format.extent | 8714628 bytes | - |
| dc.format.mimetype | application/pdf | - |
| dc.source.uri (資料來源) | http://thesis.lib.nccu.edu.tw/record/#G0111753135 | en_US |
| dc.subject (關鍵詞) | 命名資料網路 | zh_TW |
| dc.subject (關鍵詞) | 邊緣運算 | zh_TW |
| dc.subject (關鍵詞) | 發布訂閱系統 | zh_TW |
| dc.subject (關鍵詞) | 大型語言模型 | zh_TW |
| dc.subject (關鍵詞) | 自然語言處理 | zh_TW |
| dc.subject (關鍵詞) | Named Data Networking | en_US |
| dc.subject (關鍵詞) | Edge Computing | en_US |
| dc.subject (關鍵詞) | Publish/Subscribe Systems | en_US |
| dc.subject (關鍵詞) | Large Language Models | en_US |
| dc.subject (關鍵詞) | Natural Language Processing | en_US |
| dc.title (題名) | 在 NDN 發布訂閱系統中實現自然語言訂閱 | zh_TW |
| dc.title (題名) | Natural-Language Subscription over NDN Pub/Sub System | en_US |
| dc.type (資料類型) | thesis | en_US |
| dc.relation.reference (參考文獻) | [1] Patrick Th Eugster et al. “The many faces of publish/subscribe”. In: ACM comput- ing surveys (CSUR) 35.2 (2003), pp. 114–131.
[2] Kevin Chan et al. “Fuzzy interest forwarding”. In: Proceedings of the 13th Asian Internet Engineering Conference. 2017, pp. 31–37.
[3] Rishi Bommasani et al. “On the opportunities and risks of foundation models”. In: arXiv preprint arXiv:2108.07258 (2021).
[4] Apache Software Foundation. Apache Kafka Documentation. Accessed: 2023-10- 15. 2023. URL: https://kafka.apache.org/documentation/.
[5] Google Cloud. Google Cloud Pub/Sub Documentation. 2023. URL: https : / / cloud.google.com/pubsub/docs (visited on 10/15/2023).
[6] Andrew Banks and Rahul Gupta. MQTT Version 3.1.1. Tech. rep. OASIS, 2014. URL: https://docs.oasis-open.org/mqtt/mqtt/v3.1.1/mqtt-v3.1.1. html.
[7] Lixia Zhang et al. “Named data networking”. In: ACM SIGCOMM Computer Com- munication Review 44.3 (2014), pp. 66–73.
[8] Minsheng Zhang, Vince Lehman, and Lan Wang. “Scalable name-based data syn- chronization for named data networking”. In: IEEE Infocom 2017-IEEE Confer- ence on Computer Communications. IEEE. 2017, pp. 1–9.
[9] Alexander Afanasyev, Spyridon Mastorakis, et al. PSync: Partial and Full Syn- chronization Library for NDN. 2025. URL: https://github.com/named-data/ PSync.
[10] Li Yujian and Liu Bo. “A normalized Levenshtein distance metric”. In: IEEE trans- actions on pattern analysis and machine intelligence 29.6 (2007), pp. 1091–1095.
[11] Suphakit Niwattanakul et al. “Using of Jaccard coefficient for keywords similarity”. In: Proceedings of the international multiconference of engineers and computer scientists. Vol. 1. 6. 2013, pp. 380–384.
[12] Spyridon Mastorakis et al. “Experimentation with fuzzy interest forwarding in named data networking”. In: arXiv preprint arXiv:1802.03072 (2018).
[13] Adrian Zapletal, Kazuaki Ueda, and Atsushi Tagami. “Evaluation of forwarding strategies for NDN-based multi-access edge computing”. In: GLOBECOM 2020- 2020 IEEE Global Communications Conference. IEEE. 2020, pp. 1–6.
[14] JUNG Heeyoung et al. “A networking scheme for large-scale pub/sub service over NDN”. In: 2019 International Conference on Information and Communication Tech- nology Convergence (ICTC). IEEE. 2019, pp. 1195–1200.
[15] Jason Wei et al. “Chain-of-thought prompting elicits reasoning in large language models”. In: Advances in neural information processing systems 35 (2022), pp. 24824– 24837. | zh_TW |
