在今天的城市環境中，找到合適的停車位是一個重大挑戰。本研究介紹了多模態停車推薦系統（MM-PRS），旨在通過個性化的停車建議來提升使用者的滿意度。MM-PRS分為三個明確的階段。
首先，停車特徵提取（PFE）階段從使用者的停車記錄、環境數據和時間因素中提取相關特徵。這種全面的分析能深入了解使用者的偏好和停車行為模式。
其次，用戶聚類（UC）階段利用非監督機器學習技術基於這些提取特徵將用戶分組。同一群集內的用戶可從共享模型獲得量身定制的停車建議。
最後，停車場推薦（PLR）階段採用多模態技術提供定制的停車建議。這一階段包括三個集成模型：一個監督機器學習模型，用於基於使用者滿意度指標和提取特徵進行停車場的初始分類；一個兼容性過濾模型，根據使用者的距離和可用性等標準評估停車場；一個深度學習模型分析停車場和用戶特徵，以精煉和提供最佳的停車建議。
實驗結果顯示，MM-PRS在精度、召回率和F1得分方面優於現有系統。這些結果強調了MM-PRS在應對城市停車場景的複雜性方面的有效性，展示了它在顯著改善用戶停車體驗方面的潛力。未來的研究可以進一步優化算法，整合額外的影響因素，並擴展MM-PRS在實際城市環境中的應用範圍。

In today's urban environments, finding suitable parking poses a significant challenge. This study introduces the Multi-Modal Parking Recommendation System (MM-PRS), designed to enhance user satisfaction through personalized parking recommendations. MM-PRS operates in three distinct stages. Firstly, the Parking Feature Extraction (PFE) stage extracts relevant features from user parking records, environmental data, and temporal factors. This comprehensive analysis provides a deep understanding of user preferences and parking behavior patterns. Secondly, the User Clustering (UC) stage employs unsupervised machine learning techniques to group users based on these extracted features. Users within the same cluster benefit from tailored parking recommendations derived from a shared model. Lastly, the Parking Lot Recommendation (PLR) stage utilizes multi-modal techniques to offer customized parking suggestions. This stage incorporates three integrated models: a supervised machine learning model for the initial classification of parking lots based on user satisfaction metrics and extracted features, a compatibility filtering model that evaluates parking lots against user criteria such as distance and availability, and finally, a deep learning model that analyzes parking lots and user features to refine and provide optimal parking recommendations. Experimental findings highlight MM-PRS's superior performance over existing systems in terms of Precision, Recall, and F1-Score. These results underscore MM-PRS's effectiveness in addressing the complexities of urban parking scenarios, demonstrating its potential to significantly improve user parking experiences. Future research could focus on further optimizing algorithms, integrating additional influencing factors, and expanding MM-PRS applicability in practical urban settings.

Outline	V
List of Figures	VIII
List of Tables	IX
Chapter 1. Introduction	1
1. 1 Background	1
1. 2 Research Goals	3
1. 3 Organization of the Thesis	4
Chapter 2. Related Work	5
2.1 Internet of Things (IoT) Technologies	5
2.1.1 Internet of Things (IoT) architecture	5
2.1.2 Healthcare Applications	7
2.1.3 Environment Applications	9
2.1.4 Smart city Applications	10
2.1.5 Industrial Applications	11
2.1.6 Infrastructural Applications	12
2.2 Machine Learning Technologies	13
2.2.1 Classification of Machine Learning Technologies	14
2.2.2 Recommender Systems Applications	15
2.2.3 Financial Risk Management Applications	16
2.2.4 Industrial Manufacturing Applications	17
2.2.5 Personalized Education Applications	18
2.2.6 Intelligent Customer Service and Chatbots Applications	20
2.2.7 Energy Management Applications	20
2.2.8 Bioinformatics Applications	21
2.2.9 Network Security Applications	22
2.3 Deep Learning Technologies	23
2.3.1 Key components of Deep Learning Technologies	24
2.3.2 Image Recognition and Classification Applications	25
2.3.3 Natural Language Processing (NLP) Applications	26
2.3.4 Autonomous Driving Applications	27
2.3.5 Medical Diagnosis Applications	28
2.3.6 Speech Recognition and Synthesis	30
2.3.7 Generative Adversarial Networks (GANs)	31
2.4	IoT-based real-time parking space guidance	33
2.4.1 IoT-based Smart Parking Systems Overview	33
2.4.2 Limitations of Current IoT-based Systems	34
2.5	Machine learning-based parking space prediction	34
2.5.1 Enhanced Methodology and Data Preprocessing	35
2.5.2 Advantages of Enhanced Data Processing	36
2.6	Deep learning-based parking space prediction	36
Chapter 3. System Model and Problem Formulation	40
3.1 System Model and Assumptions	40
3.2 Problem Formulation	41
Chapter 4. The Proposed Mechanism	45
4.1 Parking Feature Extraction (PFE) stage	45
4.2	User Clustering (UC) stage	50
4.3	Parking Lot Recommendation (PLR) stage	52
Chapter 5. Performance Evaluation	60
5.1 Data Preparation	60
5.2	Experimental Results	61
Chapter 6. Conclusion	70
References	71

List of Figures
Fig. 2. 1. Four main components of IOT	6
Fig. 2. 2. Taxonomy of IoT applications	7
Fig. 2. 3. A typical machine learning approach.	13
Fig. 4. 1. The three stages developed in the proposed MM-PRS.	45
Fig. 4. 2. The relationship between the Cju and CjP.	52
Fig. 4. 3. The Design of the PLR phase.	54
Fig. 4. 4. The training process of Coarse-grain classification model.	56
Fig. 4. 5. The training process of fine-grain deep learning model.	59
Fig. 5. 1. The parking lots distribution of u23 in one month.	62
Fig. 5. 2. The parking behavior of u23 in one month.	62
Fig. 5. 3. The user u23 parking frequency of workdays, weekends, and holidays.	63
Fig. 5. 4. The user u23 parking features of weather and temperature.	64
Fig. 5. 5. The F1-score of MM-PRS for user u23.	65
Fig. 5. 6. Intra-Cluster and Inter-Cluster feature distance.	65
Fig. 5. 7. Comparison of MM-PRS, Coarse-Fine-grain model, and Fine-grain deep learning model by varying the days, weather, and temperature.	66
Fig. 5. 8. Comparison of MM-PRS, NCF, and DCNv2 by varying the days, weather, and temperature.	67
Fig. 5. 9. PRS adjusts the daytime parking demand at parking lots.	68
Fig. 5. 10. The performance for new users and new parking lot.	69
 
List of Tables
Table. 2. 1. The comparisons of the proposed MM-PRS and the related studies	39
Table. 4. 1. UPF-table (ui)	49

[1]	Trista. Lin, Hervé. Rivano, and Frédéric. Le Mouël. 2017. "A survey of smart parking solutions." IEEE Transactions on Intelligent Transportation Systems. 18(12): 3229-3253.
[2]	Wei. Shao, Flora D. Salim, Tao. Gu, Ngoc-Thanh. Dinh, and Jeffrey. Chan. 2018. "Traveling officer problem: Managing car parking violations efficiently using sensor data. " IEEE Internet Things Journal. 5(2): 802–810.
[3]	Seong-eun. Yoo, et al. 2010. "Guaranteeing real-time services for industrial wireless sensor networks with IEEE 802.15.4." IEEE Transactions on Industrial Electronics. 57(11): 3868-3876.
[4]	Jingyu. Liu, Jing. Wu, and Linan. Sun. 2020. "Control method of urban intelligent parking guidance system based on Internet of Things." Computer Communications. 153(Mar): 279-285.
[5]	Hongyan. Gao, et al. 2020. "Smartphone-based parking guidance algorithm and implementation." Journal of Intelligent Transportation Systems. 25(4): 412-422.
[6]	Huajun. Chai, Rui. Ma, and H. Michael Zhang. 2018. "Search for parking: A dynamic parking and route guidance system for efficient parking and traffic management." Journal of Intelligent Transportation Systems. 23(6): 541-556.
[7]	R. Lu, X. Lin, H. Zhu, and X. Shen. 2009. "SPARK: A new VANET-based smart parking scheme for large parking lots," IEEE INFOCOM 2009, Rio de Janeiro. 1413–1421. Brazil.
[8]	Thompson, Russell G, Kunimichi Takada, and Saturo Kobayakawa. 2001. "Optimisation of parking guidance and information systems display configurations." Transportation Research Part C: Emerging Technologies.  9(1): 69-85.
[9]	Amir O. Kotb, Yao-chun. Shen and Yi. Huang. 2017. "Smart parking guidance, monitoring, and reservations: A Review," IEEE Intelligent Transportation Systems Magazine. 9(2): 6-16.
[10]	Arun. Varghese, and G. Sreelekha. 2019. "An efficient algorithm for detection of vacant spaces in delimited and non-delimited parking lots." IEEE Transactions on Intelligent Transportation Systems. 21(10): 4052-4062.
[11]	Xiaofei. Ye, Jinfen. Wang, Tao. Xingchen. Yan, Qiming. Ye and Jun. Chen. 2020., "Short-term prediction of available parking space based on machine learning approaches." IEEE Access. 8(Sep): 174530-174541.
[12]	Shuguan Yang, Wein. Ma, Xidong, Pi, and Sean Qian. 2019. "A deep learning approach to real-time parking occupancy prediction in transportation networks incorporating multiple spatio-temporal data sources." Transportation Research Part C: Emerging Technologies. 107(Oct): 248-265.
[13]	Yonghong. Liu, Chunyu. Liu, and Xia. Luo. 2021. "Spatiotemporal deep-learning networks for shared-parking demand prediction." Journal of Transportation Engineering, Part A: Systems. 147(6): 04021026.
[14]	Jun. Li, Haohao. Qu and Linlin. You. 2023. "An integrated approach for the near real-time parking occupancy prediction," IEEE Transactions on Intelligent Transportation Systems. 24(4): 3769-3778.
[15]	Eleni I. Vlahogianni, Konstantinos. Kepaptsoglou, Vassileios. Tsetsos, and Matthew G. Karlaftis. 2015. "A real-time parking prediction system for smart cities." Journal of Intelligent Transportation Systems. 20(2): 192–204.
[16]	Jae Kyu. Suhr and Ho Gi. Jung. 2013. "Sensor fusion-based vacant parking slot detection and tracking." IEEE Transactions on Intelligent Transportation Systems.15(1): 21-36.
[17]	Zhihua. Cui, et al. 2020. "Personalized recommendation system based on collaborative filtering for IoT scenarios," IEEE Transactions on Services Computing. 13(4): 685-695.
[18]	Hassan R, Qamar F, Hasan M K, et al. 2020. "Internet of Things and its applications: A comprehensive survey". Symmetry, 12(10): 1674.
[19]	Guoping. Zeng. 2019. "On the confusion matrix in credit scoring and its analytical properties," Communications in Statistics-theory and Methods. 49(9): 2080-2093.
[20]	Wu. Wei, Jun. Yan, Xiaofu. Wu, and Chen. Wang. 2021. "A data preprocessing method for deep learning-based device-free localization." IEEE Communications Letters. 25(12): 3868-3872.
[21]	Chang. Liu, Liu. Yang, and jingyi. Qu. 2021. "A structured data preprocessing method based on hybrid encoding." Journal of Physics: Conference Series. 1738(1): 12060.
[22]	Shichao. Zhang, Chengqi. Zhang, and Qiang. Yang. 2010. "Data preparation for data mining." Applied artificial intelligence. 17(5-6): 375-381.
[23]	Hari Prasanna. Das and  Costas J. Spanos. 2022. "Improved dequantization and normalization methods for tabular data pre-processing in smart buildings." Proceedings of the 9th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation. 168-177.
[24]	Dalwinder. Singh and Birmohan. Singh. 2020."Investigating the impact of data normalization on classification performance." Applied Soft Computing. 97(B): 105524.
[25]	Shyr-Shen. Yu, et al. 2018. "Two improved k-means algorithms." Applied Soft Computing.  68(Jul): 747-755.
[26]	T. Kanungo, et al. 2002. "An efficient k-means clustering algorithm: analysis and implementation." IEEE transactions on pattern analysis and machine intelligence.   24(7): 881-892.
[27]	Yiping. Gan, et al. 2023. "Prediction of progressive collapse resistance of RC frames using deep and cross network model." Structures. 51(May): 800-813.
[28]	Ruoxi. Wang, et al. 2021. "Dcn v2: Improved deep & cross network and practical lessons for web-scale learning to rank systems." Proceedings of the web conference. 1785-1797.
[29]	Khaki. Saeed, Zahra. Khalilzadeh, and Lizhi. Wang. 2020. "Predicting yield performance of parents in plant breeding: A neural collaborative filtering approach." Plos one. 15(5): e0233382.
[30]	Pierre. De Handschutter, Nicolas. Gillis, and Xavier. Siebert. 2021. "A survey on deep matrix factorizations." Computer Science Review. 42(Nov): 100423.