HMM-Based Map Matching and Spatiotemporal Analysis for Matching Errors with Taxi Trajectories

Round 1

Reviewer 1 Report

This article focuses on the problem of map-matching using HMM. The authors identified four types of map-matching errors. They used a dataset with taxi trajectories fin Chengdu, China. They also analyzed spatio-temporal patterns of map-matching errors and the contributing factors related to different errors likehoods. The article is well written and easy to follow. The map-matching problem is a hot topic and I believe it is of interest of IJGI readers. I have the following concerns concerning the article:

1) I would like to see a Related Work section, in which the authors could discuss SOTA in map-matching and compare the different approaches.

2) It should be clear what are the contributions of the proposed method when compared do SOTA.

3) Why choosing HMM instead of Deep learning models? Please justify this decision.

4) Why did you choose the four types of map-matching errors? What are other possible errors?

5) It would be interesting do discuss on the performance metrics  (CPU time, memory size, disk size) of the proposed approach.

6) Several important references are missing such as:

Kanta Prasad Sharma, Ramesh Chandra Poonia, Surendra Sunda: A novel map matching algorithm for real-time location using low frequency floating trajectory data.  Int. J. Adv. Intell. Paradigms 24(3/4): 442-455 (2023).

Jie Fang, Xiongwei Wu, DianChao Lin, Mengyun Xu, Huahua Wu, Xuesong Wu, Ting Bi: A Map-Matching Algorithm With Extraction of Multi-Group Information for Low-Frequency Data.  IEEE Intell. Transp. Syst. Mag. 15(2): 238-250 (2023)

Wanting Li, Yongcai Wang, Deying Li, Xiaojia Xu: A robust map matching method by considering memorized multiple matching candidates.  Theor. Comput. Sci. 941: 104-120 (2023)

Hanwen Hu, Shiyou Qian, Jingchao Ouyang, Jian Cao, Han Han, Jie Wang, Yirong Chen: AMM: An Adaptive Online Map Matching Algorithm.  IEEE Trans. Intell. Transp. Syst. 24(5): 5039-5051 (2023)

Linli Jiang, Chaoxiong Chen, Chao Chen: L2MM: Learning to Map Matching with Deep Models for Low-Quality GPS Trajectory Data.  ACM Trans. Knowl. Discov. Data 17(3): 39:1-39:25 (2023)

Shengjie Ma, Hyukjoon Lee: A Practical HMM-Based Map-Matching Method for Pedestrian Navigation. ICOIN 2023: 806-811

Matteo Tortora, Ermanno Cordelli, Paolo Soda: PyTrack: A Map-Matching-Based Python Toolbox for Vehicle Trajectory Reconstruction. IEEE Access 10: 112713-112720 (2022

Subhrasankha Dey, Martin Tomko, Stephan Winter: Map-Matching Error Identification in the Absence of Ground Truth. ISPRS Int. J. Geo Inf. 11(11): 538 (2022)

Siavash Saki, Tobias Hagen: A Practical Guide to an Open-Source Map-Matching Approach for Big GPS Data. SN Comput. Sci. 3(5): 415 (2022)

Marko Dogramadzi, Aftab Khan: Accelerated Map Matching for GPS Trajectories. IEEE Trans. Intell. Transp. Syst. 23(5): 4593-4602 (2022)

Qingying Yu, Fan Hu, Zhen Ye, Chuanming Chen, Liping Sun, Yonglong Luo: High-Frequency Trajectory Map Matching Algorithm Based on Road Network Topology.  IEEE Trans. Intell. Transp. Syst. 23(10): 17530-17545 (2022)

Jie Feng, Yong Li, Kai Zhao, Zhao Xu, Tong Xia, Jinglin Zhang, Depeng Jin: DeepMM: Deep Learning Based Map Matching With Data Augmentation.  IEEE Trans. Mob. Comput. 21(7): 2372-2384 (2022)

Zhenfeng Huang, Shaojie Qiao, Nan Han, Chang-an Yuan, Xuejiang Song, Yueqiang Xiao: Survey on vehicle map matching techniques. CAAI Trans. Intell. Technol. 6(1): 55-71 (2021)

 

Author Response

Thanks for providing us with this great opportunity to submit a revised version of our manuscript. We appreciate the detailed and constructive comments provided by the reviewers, and we have carefully revised the manuscript by incorporating all the suggestions of the review panel.

We would like to take this opportunity to thank you for all your time involved and for this great opportunity for us to improve the manuscript. We hope you will find this revised version satisfactory.

Author Response File: Author Response.docx

Reviewer 2 Report

1. The abstract mentions that "few literatures focus on exploring map-matching errors." It would be helpful to provide a more detailed review of existing research on map-matching errors and explain how your study builds upon and contributes to this body of literature. This would better contextualize your research and demonstrate its novelty.

2. The methodology section should provide a clearer explanation of the HMM algorithm and how it is used for map matching. Additionally, please provide more details on the criteria used to classify the four types of map-matching errors. It would be beneficial for readers to understand the rationale behind these classifications.

3. Please provide more information on the data sources used in the study, particularly the taxi trajectories and roadway network data. Discuss any preprocessing or data cleaning steps performed on the data to ensure its accuracy and reliability.

4. In the conclusion section, authors mention several limitations of the study. It would be helpful to discuss how these limitations could be addressed in future research. This could include exploring alternative error classification methods, incorporating additional factors such as traffic conditions and driver behavior, or using more accurate trajectory data sources.

The quality of the English language in the manuscript is acceptable. However, there are several areas that require improvement to enhance the clarity and readability of the text. I recommend that the authors carefully proofread the manuscript.

Author Response

Thanks for providing us with this great opportunity to submit a revised version of our manuscript. We appreciate the detailed and constructive comments provided by the reviewers, and we have carefully revised the manuscript by incorporating all the suggestions of the review panel.

We would like to take this opportunity to thank you for all your time involved and for this great opportunity for us to improve the manuscript. We hope you will find this revised version satisfactory.

Author Response File: Author Response.docx

Reviewer 3 Report

This study proposes a Hidden Markov Model (HMM) algorithm to match trajectories on road networks, and a temporal Kernel density analysis with a multinomial logistic model to examine the spatial-temporal patterns of four kinds of trajectories map-matching errors (Off-Road Error [ORE], Wrong-match on Road Error [WRE], Off-Junction Error [OJE], and Wrong-match in Junction Error [WJR]).

The analysis was applied on a dataset with trajectories derived from GPS data collected from taxis, and obtained from Chengdu Municipal Traffic Management Bureau in China, from September 1st to 14th of 2020 (14M trajectories generated by 500 randomly selected taxis).

From the analysis several key findings are derived about when or where the probability of a map-matching error is higher, or other impacts (like intersection features, road characteristics, speed conditions etc.)

 

Further comments/suggestions/improvements:

A. A strong motivation for this study is to improve the effectiveness of map matching for commercial vehicles. However, important factors for such an improvement are not provided:

i. How much time is required in order to detect and correct a map-matching error with the proposed method?

ii. Can this response time be supported by an existing on-line map matching system?

iii. Is there any preprocessing phase that is required for the detection?

* Further experimental results are required to study these factors and support the motivation

 

B. Important technical and implementation details for the proposed methodology are not provided:

i. Detailed procedure/algorithm for the preprocessing phase.

ii. Detailed procedure/algorithm for HMM, Kernel density analysis and final map-matching error detection.

iii. What data structures were used to hold the required data and all intermediate calculations?

iv. What is the total spatial and temporal complexity for the proposed method?

v. In what kind of computer systems, the experiments were conducted?

vi. What software/libraries/languages have been used for the implementation of the proposed method?

 

Author Response

Thanks for providing us with this great opportunity to submit a revised version of our manuscript. We appreciate the detailed and constructive comments provided by the reviewers, and we have carefully revised the manuscript by incorporating all the suggestions of the review panel.

We would like to take this opportunity to thank you for all your time involved and for this great opportunity for us to improve the manuscript. We hope you will find this revised version satisfactory.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

I would like to thank the authors for incorporating the issues I demanded in the first round. Hence, I am favourable to publish this article in the IJGI.

The authors improved English quality in the paper.

Author Response

Thanks you for your positive comments on our work. We have diligently polished the overall language of the manuscript. 

Reviewer 2 Report

1. You have added the processing time for identifying map matching errors, which took 4,550 seconds total for 14 million GPS points. An online map matching system is not considered currently but can be explored in future work.

2. Explanations were added on the choice of the four error types, limitations, and future work directions. 

3. More context was given on existing research on map matching errors and how this work contributes. The HMM algorithm and error classification criteria are explained clearer. 

4. Additional details were provided on the data sources, preprocessing, and limitations. 

These replies completely solved the opinions of the reviewers and significantly improved the manuscript. The new content provides better motivation, technical details, literature analysis and future work. In my opinion, the revised paper should be suitable for publication after incorporating these amendments.

Author Response

Thanks you for your postive statement on our work!

Reviewer 3 Report

In the revised version, the authors 

successfully addressed all mentioned 

comments with signifficant effort. 

They provided additional details,

references, and important information 

for the proposed methodology.

Moreover, they added more technical 

and implementation details, and more 

results. 

Author Response

Thank you!