近年來，人工智慧技術快速發展改變了資訊檢索的方式，而嵌入空間已成為現代人工智慧模型的重要資料表示方式。因此，如何在嵌入空間中支援探索式搜尋是為一項重要的研究課題。然而，現有的探索式搜尋系統普遍提供有限的互動機制，且未能明確建立於既有的探索式搜尋理論。本論文提出一個適用於嵌入空間的動態互動式探索搜尋框架，並以影像探索系統作為實作驗證。受 漿果採摘模型 (Berry-Picking Model)、資訊覓食理論 (Information Foraging Theory) 與地圖導航隱喻啟發，本研究定義了四種搜尋模式，包括細部搜尋（Detailed Search）、延伸搜尋（Extended Search）、方向－距離搜尋（Direction–Distance Search）以及過渡搜尋（Transition Search）。為支援上述互動方式，本研究提出一種整合階層式分群與群集相似性圖的樹－圖資料結構。此外，本研究使用一種基於編碼器的參數化降維方法，同時最佳化歐氏距離與餘弦相似度，以保留嵌入空間中的幾何關係。基於上述框架，本研究設計並實作四種視覺化以作使用者介面，包括謝爾賓斯基方形樹狀視圖 (Sierpiński Square Tree View)、地形密度視圖 (Topographic Density View)、方向－距離視圖 (Direction–Distance View) 及景觀視圖 (Landscape View)。該系統利用 Flower Recognition 和 WikiArt 資料集，結合 SigLIP 和 DINOv2 影像嵌入技術進行了實現，並評估所提出的降維方法及景觀視圖策略的有效性。最後，透過案例研究驗證本系統能有效支援知識獲取、語意轉換探索、異常偵測、嵌入空間分析，以及多元推薦發掘等探索式搜尋任務。研究結果顯示，本論文所提出之框架能為高維嵌入空間中的互動式探索搜尋提供一套兼具理論基礎與實用性的方法。

The rapid advancement of artificial intelligence has transformed information retrieval, and embedding spaces have become a fundamental representation of modern AI models. Supporting exploratory search in embedding spaces has become an important challenge. However, existing exploratory search systems often provide limited interaction mechanisms and lack explicit grounding in established exploratory search theories. This thesis presents a dynamic interactive exploratory search framework for embedding spaces, demonstrated through an image exploration system. Inspired by the Berry-Picking Model, Information Foraging Theory, and map navigation metaphors, we define four search modes: detailed search, extended search, direction–distance search, and transition search. To support interactions, we propose a tree–graph data structure that integrates hierarchical clustering with a cluster similarity graph. We further introduce an encoder-based parametric dimensionality reduction method that jointly optimizes Euclidean distance and cosine similarity to preserve geometric relationships. Based on this framework, we develop four visualization interfaces: Sierpiński Square Tree View, Topographic Density View, Direction–Distance View, and Landscape View. The system is implemented using the Flower Recognition and WikiArt datasets with SigLIP and DINOv2 image embeddings to evaluate the proposed dimensionality reduction method and the Landscape View strategy. Finally, use case studies demonstrate that the system supports knowledge acquisition, semantic transition exploration, anomaly detection, embedding analysis, and diverse recommendation discovery. These results demonstrate that the proposed framework provides an effective and theoretically grounded approach for interactive exploratory search in high-dimensional embedding spaces.

[1] T. Karunaratne and A. Adesina, “Is it the new google: Impact of chatgpt on students’ information search habits,” in Proceedings of the 22nd European Conference on e-Learning, ECEL, 2023, pp. 147–155.
[2] C. W. Choo, B. Detlor, and D. Turnbull, “Information seeking on the web: An integrated model of browsing and searching,” First Monday, vol. 5, no. 2, Feb. 2000. [Online]. Available: https://firstmonday.org/ojs/index.php/fm/article/view/729
[3] X. Zhang, H. Kang, Y. Cai, and T. Jia, “Clip model for images to textual prompts based on top-k neighbors,” in 2023 3rd International Conference on Electronic In-formation Engineering and Computer Science (EIECS), 2023, pp. 821–824.
[4] B. Sun, P. Zhou, L. Du, and X. Li, “Active deep image clustering,” Knowledge-Based Systems, vol. 252, p. 109346, 2022. [Online]. Available: https://www. sciencedirect.com/science/article/pii/S095070512200675X
[5] Z. Zhu and K. Mao, “Knowledge-based bert word embedding fine-tuning for emotion recognition,” Neurocomputing, vol. 552, p. 126488, 2023. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S0925231223006112
[6] R. W. White, B. Kules, and B. Bederson, “Exploratory search interfaces: categorization, clustering and beyond: report on the xsi 2005 workshop at the human-computer interaction laboratory, university of maryland,” SIGIR Forum, vol. 39, no. 2, p. 52–56, Dec. 2005. [Online]. Available: https://doi.org/10.1145/1113343.1113356
[7] G. Marchionini, “Exploratory search: from finding to understanding,” Commun. ACM, vol. 49, no. 4, p. 41–46, Apr. 2006. [Online]. Available: https://doi.org/10.1145/1121949.1121979
[8] E. Palagi, F. Gandon, A. Giboin, and R. Troncy, “A survey of definitions and models of exploratory search,” in Proceedings of the 2017 ACM Workshop on Exploratory Search and Interactive Data Analytics, ser. ESIDA ’17. New York, NY, USA: Association for Computing Machinery, 2017, p. 3–8. [Online]. Available: https://doi.org/10.1145/3038462.3038465
[9] M. J. Bates, “The design of browsing and berrypicking techniques for the online search interface,” Online Review, vol. 13, no. 5, pp. 407–424, 05 1989. [Online]. Available: https://doi.org/10.1108/eb024320
[10] P. Pirolli and S. Card, “Information foraging.” Psychological Review, vol. 106, pp. 643–675, 1999.
[11] B. Kules, R. Capra, M. Banta, and T. Sierra, “What do exploratory searchers look at in a faceted search interface?” in Proceedings of the 9th ACM/IEEE-CS Joint Conference on Digital Libraries, ser. JCDL ’09. New York, NY, USA: Association for Computing Machinery, 2009, p. 313–322. [Online]. Available: https://doi.org/10.1145/1555400.1555452
[12] C. D. Sciascio, V. Sabol, and E. Veas, “Supporting exploratory search with a visual user-driven approach,” ACM Trans. Interact. Intell. Syst., vol. 7, no. 4, Dec. 2017. [Online]. Available: https://doi.org/10.1145/3009976
[13] K. Verbert, D. Parra, P. Brusilovsky, and E. Duval, “Visualizing recommendations to support exploration, transparency and controllability,” in Proceedings of the 2013 International Conference on Intelligent User Interfaces, ser. IUI ’13. New York, NY, USA: Association for Computing Machinery, 2013, p. 351–362. [Online]. Available: https://doi.org/10.1145/2449396.2449442
[14] T. Ruotsalo, J. Peltonen, M. Eugster, D. Głowacka, K. Konyushkova, K. Athukorala, I. Kosunen, A. Reijonen, P. Myllymäki, G. Jacucci, and S. Kaski, “Directing exploratory search with interactive intent modeling,” in Proceedings of the 22nd ACM International Conference on Information & Knowledge Management, ser. CIKM ’13. New York, NY, USA: Association for Computing Machinery, 2013, p. 1759–1764. [Online]. Available: https://doi.org/10.1145/2505515.2505644
[15] C. Seifert, J. Jurgovsky, and M. Granitzer, “Facetscape: A visualization for exploring the search space,” in 2014 18th International Conference on Information Visualisa-tion, 2014, pp. 94–101.
[16] J. wook Ahn and P. Brusilovsky, “Adaptive visualization for exploratory information retrieval,” Information Processing Management, vol. 49, no. 5, pp. 1139–1164, 2013. [Online]. Available: https://www.sciencedirect.com/science/ article/pii/S0306457313000137
[17] D. Parra, P. Brusilovsky, and C. Trattner, “See what you want to see: visual user-driven approach for hybrid recommendation,” in Proceedings of the 19th International Conference on Intelligent User Interfaces, ser. IUI ’14. New York, NY, USA: Association for Computing Machinery, 2014, p. 235–240. [Online]. Available: https://doi.org/10.1145/2557500.2557542
[18] D. Bertucci, M. M. Hamid, Y. Anand, A. Ruangrotsakun, D. Tabatabai, M. Perez, and M. Kahng, “Dendromap: Visual exploration of large-scale image datasets for machine learning with treemaps,” IEEE Transactions on Visualization and Computer Graphics, vol. 29, no. 1, pp. 320–330, 2023.
[19] A. Bäuerle, C. v. Onzenoodt, D. Jönsson, and T. Ropinski, “Semantic Hierarchical Exploration of Large Image Datasets,” in EuroVis 2023 - Short Papers, T. Hoellt, W. Aigner, and B. Wang, Eds. The Eurographics Association, 2023.
[20] Z. J. Wang, F. Hohman, and D. H. Chau, “WizMap: Scalable interactive visualization for exploring large machine learning embeddings,” in Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 3: System Demonstrations), D. Bollegala, R. Huang, and A. Ritter, Eds. Toronto, Canada: Association for Computational Linguistics, Jul. 2023, pp. 516–523. [Online]. Available: https://aclanthology.org/2023.acl-demo.50/
[21] C.-H. Lai, M.-F. Kuo, Y.-H. Lien, K.-A. Su, and Y.-S. Wang, “Parametric dimen-sion reduction by preserving local structure,” in 2022 IEEE Visualization and Visual Analytics (VIS), 2022, pp. 75–79.
[22] T. Fujiwara, R. Matsushita, M. Iwamaru, M. Tange, S. Someya, and K. Okamoto, “Fractal map: Fractal-based 2d expansion method for multi-scale high-dimensional data visualization,” in Advances in Visual Computing: 6th International Symposium, ISVC 2010, Las Vegas, NV, USA, November 29-December 1, 2010. Proceedings, Part I. Berlin, Heidelberg: Springer-Verlag, 2010, p. 306–315. [Online]. Available: https://doi.org/10.1007/978-3-642-17289-2_30