A Method for Recognizing I-Shaped Building Patterns Utilizing Multi-Scale Data and Knowledge Graph

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper presents a novel and effective hybrid approach for a specific recognition task, i.e., I-shaped building patterns. However, there are several weak points as follows:

1. the approach relies on constructing a specific knowledge graph with "cross-scale recognition rules" for the I-shaped pattern. The effort required to define these rules for other, more complex building patterns may be significant, potentially limiting the method's scalability to a broader set of urban patterns. The scalability and generality should be discussed.
2. While effective for a well-defined pattern like "I-shaped," rule-based systems can be brittle. They may struggle with ambiguous cases, noisy data, or patterns that require more nuanced, probabilistic interpretation. The performance is heavily dependent on the completeness and correctness of the manually encoded rules. Additional experiments should be conducted to show the robustness of the method.
3. The dataset in the experiments is limited, and only two scale levels are considered, which is insufficient to demonstrate the merit of the proposed method in this paper on multi-scale data.
4. There are also some typos in the paper:

• Line-203: the subsection title “2 Building Pattern Recognition”is the same with 2.1(Line-143).
• Inappropriate use of spaces, such as Line-51 “building  features”, Line-72 “patterns.Nonetheless”, etc.
• Inappropriate use of mathematical symbols, such as Line-300 “θ1, θ2, θ3”, where the numbers should be in lowercase, i.e., “θ₁, θ₂, θ₃”. Also the same problem in the equations.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This paper uses knowledge graph rule reasoning and multi-scale feature recognition to identify I-shaped building patterns, and the method is relatively novel. The following comments are provided:

1. It is necessary to further explain the value of building pattern recognition. For example, how identifying I-shaped, C-shaped, or E-shaped building layouts can "better identify the current bottlenecks in urban development and directions for future expansion" (Line 44).
2. The title "2.2. Building Pattern Recognition" is duplicated with "2.1. Building Pattern Recognition".
3. Line 455: "assess the relationships of these recognized buildings with structures at other scales, either smaller or larger, to determine the most suitable enhancement recognition approach": how is this "determination" made? Is the process automatic?
4. In the two experimental datasets, how were the Gold-standard I-shaped building patterns obtained? What is the number of real I-shaped building patterns in each dataset?
5. Line 542: "template matching method (TM)": What is the source of the template (constructed by you or derived from other articles)? What is its content?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Line 78–81: The authors state that 1(a) and 1(c) can be identified by existing methodologies, whereas 1(b) and 1(d) cannot. This statement is somewhat confusing. Since 1(a) and 1(b) represent simpler patterns compared to 1(c) and 1(d)—which contain more small-scale components—it is unclear why the current methods succeed on 1(a) and 1(c) but fail on 1(b) and 1(d). If this observation is accurate, please provide relevant references or supporting studies to substantiate the claim.

Sections 2.1 and 2.2: Both sections share the same subtitle. It is recommended to use distinct subtitles to highlight the central topics or themes discussed in each part of the literature review.

Line 269: Why are there specifically 13 fundamental relative positions? Provide justification or references that support this number.

Line 298: The explanation of Equation (2) is unclear—particularly, why is there a subtraction term Ab – 1? Please elaborate on its mathematical or conceptual meaning.

Lines 303–307: Add more explanation for Equation (3). For example, explain how this relationship changes (PerO change) when Oa and Ob are perpendicular versus non-perpendicular.

Line 315: Is the distance between A and B₁ measured using the centroids of A and B₁? Please add a clearer description of this formula and justify the subtraction of 1 in the expression.

Line 327: The text refers to a “Man-to-Many (M:M)” relationship, but Figure 5 shows “M:N.” Please ensure consistency between the text and figure.

Line 335: Should this refer to Equation (6) or Equation (5)? Please verify and correct.

Line 337: The threshold value of 0.3 requires justification. Is this threshold supported by existing studies or based on professional standards? Please cite relevant sources or explain the rationale for choosing this value.

Line 470: The study areas in Lanzhou and London appear to cover only portions of the cities rather than their entire extents. Please include inset maps clearly indicating where the study areas are located within the broader contexts of Lanzhou and London to enhance spatial clarity.

Line 514: In the map, is the result of shape recognition (shown in blue) explicitly displayed? Please clarify to make it more interpretable.

Line 521: Verify the figure numbering—is this referring to Figure 12?

Lines 524–525: The sentence “This approach not only identifies I-shaped patterns at specified scales but also amplifies the identification process for detecting potential I-shaped configurations across varying scales” needs further clarification. How does the proposed approach identify I-shaped patterns? Figure 12 only shows BuildingBig, BuildingSmall, Neighbor, and Match. Please provide additional visual or textual explanation to demonstrate how the I-shaped configurations are identified.

Line 555: Should the total numbers of Tr, Fr, and Mr be consistent across all methods? For example, in SR+AR, the totals (38 + 11 + 12 = 61) differ from SR+AR+CR (38 + 12 = 50). Please check the calculations and ensure consistency. The same issue applies to the following methods.

Line 580: Again, please verify whether this reference to Figure 10 is correct.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

This study proposed a knowledge-graph-based approach to recognizing I-shaped building patterns from multi-scale datasets of Lanzhou and London. Although the topic is interesting, several issues still need to be addressed.

1. In the abstract, please clarify the methodology used to extract building relationships, as it is not clearly stated. The current description leaves readers uncertain about how the relationships were derived.
2. In the introduction (page 2, line 60), please provide stronger justification and supporting references to substantiate the argument. Otherwise, it appears more like the authors’ assumption rather than a widely supported fact.
3. Regarding the issue of misrecognizing buildings in “one-to-many” or “many-to-one” cases, this study's analysis is limited only to I-shaped buildings. Other common shapes, such as U-shaped, L-shaped, and O-shaped buildings, are widely found in real-world datasets and should also be considered. Including these forms would strengthen the comprehensiveness and generalizability of the proposed approach.
4. In Section 2, instead of presenting a simple list of previous studies, it is recommended to create a summary table showing key concepts, strengths, weaknesses, limitations, etc. This will better highlight the room for improvement and the advantages of applying the knowledge-graph approach.
5. Additionally, the literature review appears to miss recent work on building shape recognition, such as From Footprints to Functions: A Comprehensive Global and Semantic Building Footprint Dataset; Deep Learning Based Urban Building Coverage Ratio Estimation Focusing on Rapid Urbanization Areas; and Deep Learning-Based Building Footprint Extraction With Missing Annotations. These advanced approaches should be discussed to better position the contribution of this study.
6. Overall, both the introduction and literature review sections are lengthy and somewhat repetitive. Streamlining these sections would improve clarity and focus.
7. In Section 3, it would be helpful to discuss whether a deep learning approach with sufficient labeled data could also address the research problem. Please clarify the comparative advantages of the proposed KG-based method over DL-based alternatives.
8. On page 8, line 278, the authors stated that the building layouts in the red box are uncommon. However, some of the excluded configurations (especially attached buildings at or near corners) are not rare in real life and can be observed in datasets such as Lanzhou and London, as shown in Figure 8.
9. On page 19, line 546, the results were validated through visual inspection by seven graduate students. This validation method should be improved. Since the study uses the National Basic Geographic Information Database (cadastral data), does this database include building union information that could serve as ground truth? Using such authoritative data would provide a more rigorous validation of the proposed KG-based method.
10. There are many typo errors, which must be revised (e.g., page 3, line 88; page 5, line 186)

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 5 Report

Comments and Suggestions for Authors

The article introduces a method for the automatic recognition of I-shaped buildings. The topic discussed has a double interest connected to both a study on urban planning and city development, that for the geometric implications related to that shape.

The structure of the article is clearly defined at the end of the Introduction section. The contents and the aims of all the paragraphs are proposed by the Authors. The section “Related Works” proposes a wide analysis of the literature on this topic, highlighting the recognition techniques used and proposing a critical overview of other research. Please note that sections 2.1 and 2.2 have the same title. Sections 3 and 4 define and describe the method used for the construction of the knowledge graph. A wide use of images in this section clearly shows the approach followed and helps the readers in the comprehension of the text. It could be interesting to describe the geometric constraints that characterize the 13 fundamental relative positions for each axis. The “Experiments and Analysis” section describes the application to a case study. Interesting the comparison between the results of the automatic detection of the shape and the visual recognition. The “Discussion” section at the end clearly analyzes the results, opening a critical evaluation of the multi-scale approach.

References are adequate.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Although the response has been well described in the coverletter, the submitted version of the paper is still the old one. Please submit the right version!

Author Response

Comment 1：Although the response has been well described in the coverletter, the submitted version of the paper is still the old one. Please submit the right version!

Response 1：Thank you very much for pointing this out. We sincerely apologize for this oversight. We have now carefully checked the submission in the online system and uploaded the correct revised version as an attachment, which fully incorporates all the responses and modifications described in the cover letter. We appreciate the reviewer’s patience and apologize for any inconvenience caused.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have carefully revised the manuscript according to the comments.

Author Response

Comment ：The authors have carefully revised the manuscript according to the comments.

Response ：We sincerely thank the reviewer for the positive evaluation and for acknowledging our revisions. We greatly appreciate the reviewer’s time and constructive feedback, which have helped us improve the quality and clarity of the manuscript.

Reviewer 4 Report

Comments and Suggestions for Authors

The authors have addressed most of my previous comments. In this revised manuscript, Table 2 should be revised to maintain description style consistency in the column "Representative studies" as well as enhance the layout design by adding borders.

Author Response

Comment : The authors have addressed most of my previous comments. In this revised manuscript, Table 2 should be revised to maintain description style consistency in the column "Representative studies" as well as enhance the layout design by adding borders.

Response : Thank you very much for the reviewer’s careful review and helpful suggestion. We have revised Table 1 and Table 2 in the manuscript to ensure a consistent description style in the column “Representative studies”. In addition, the table layout has been improved by adding borders to enhance readability and visual clarity. These revisions have been incorporated into the revised manuscript. Please refer to the attachment for the modification effect of the table.

Author Response File: Author Response.pdf