type:: [[REVIEWS]]
tags:: 
year:: 2024
venue:: [[MODELS-DS]] 
full-title:: Ensuring the consistency of information between two versions of a mechanical drawing
date-start:: [[23-07-2024]] - 17:32
date-submitted::
external-links::
status:: [[DONE]] 
deadline-submission:: 
file:: [[@Ensuring the consistency of information between two versions of a mechanical drawing]] 
parent::
todoist:: https://app.todoist.com/app/task/9-alexandre-monnier-weil-ensuring-the-consistency-of-information-between-two-ver-6Vm6Gv567J6F2gJ6

	- ### [[Highlights]]
		- # Annotazioni
			- (23/7/2024, 17:32:02)
			- - “Companies face a recurring issue with managing their hand-drawn drawings produced in the mid-20th century” ([Weil, 2017, p. 1](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=1&annotation=LR9HALXE)) #5fb236
			- - “This transformation process must be error-free to avoid material waste and significant costs.” ([Weil, 2017, p. 1](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=1&annotation=IUTNQRI7)) #5fb236
			- - “This verification process is time-consuming and laborious, often requiring the dedicated effort of two engineers for an entire week per drawing” ([Weil, 2017, p. 1](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=1&annotation=FKU5UJ6B)) #5fb236
			- - “Identifying these dissimilarities will reduce the effort of the engineers to perform the necessary corrections” ([Weil, 2017, p. 1](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=1&annotation=4VDJNQAN)) #a28ae5
			- - “Our approach is divided into two aspects:” ([Weil, 2017, p. 1](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=1&annotation=KAH49YUK)) #5fb236
			- - “interpreting the drawings to extract information” ([Weil, 2017, p. 1](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=1&annotation=SIABIJTL)) #a28ae5
			- - “comparing the drawings based on this information” ([Weil, 2017, p. 1](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=1&annotation=U733WWMA)) #a28ae5
			- - “formalism to describe the information contained in mechanical drawings” ([Weil, 2017, p. 1](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=1&annotation=3EBIEYYV)) #5fb236
			- - “graph-based structure” ([Weil, 2017, p. 1](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=1&annotation=QKWQ64EN)) #5fb236
			- - “Figure 2.” ([Weil, 2017, p. 1](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=1&annotation=J77ZISZR)) #ffd400
			- *This should come before Fig. 1, isn't it?*
			- - “Any mistake can propagate through the production process, resulting in material waste and significant costs” ([Weil, 2017, p. 1](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=1&annotation=I4VGH8HM)) #a28ae5
			- - “This verification process is time-consuming and laborious, often requiring the dedicated effort of two engineers for an entire week per drawing” ([Weil, 2017, p. 1](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=1&annotation=8XC97Z3Q)) #a28ae5
			- - “automate this verification step;” ([Weil, 2017, p. 1](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=1&annotation=QMP3UHDB)) #5fb236
			- - “dissimilarity as a divergence of information that leads to inconsistencies in the CAD drawing with respect to the original hand-drawn version.” ([Weil, 2017, p. 1](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=1&annotation=SNYJHL53)) #a28ae5
			- - “Identifying these dissimilarities will reduce the effort of the engineers to perform the necessary corrections.” ([Weil, 2017, p. 1](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=1&annotation=I6M4RJYF)) #a28ae5
			- - “the placement of annotations is different between the versions” ([Weil, 2017, p. 2](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=2&annotation=E6GRQM28)) #ffd400
			- *Does this represent a problem?*
			- - “For example, in Figure 2, the dimension values are not in the same location, and the arrowheads in the hand-drawn drawing point outward, while in the CAD drawing, they point inward.” ([Weil, 2017, p. 2](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=2&annotation=XYLWPKGW)) #ffd400
			- *There are also some values that are different.*
			- - “This complexity shows that an interpretation step of the drawings is necessary before considering comparison.” ([Weil, 2017, p. 2](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=2&annotation=YH4UZSYT)) #a28ae5
			- - “Graph construction” ([Weil, 2017, p. 2](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=2&annotation=KFT22CXD)) #2ea8e5
			- - “Drawing interpretation” ([Weil, 2017, p. 2](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=2&annotation=AID8UUTQ)) #2ea8e5
			- - “developing a methodology for segmenting annotations and pure geometry” ([Weil, 2017, p. 2](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=2&annotation=KN928MSC)) #5fb236
			- - “incrementally adding new examples to enhance annotation detection.” ([Weil, 2017, p. 2](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=2&annotation=FTUA7WAV)) #a28ae5
			- - “Interpretation-Based comparison” ([Weil, 2017, p. 2](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=2&annotation=MSV6GA6N)) #2ea8e5
			- - “Integrated methodology validation” ([Weil, 2017, p. 2](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=2&annotation=RM3MZJQ5)) #2ea8e5
			- - “scanned hand-drawn drawings are algorithmically cleaned of all unwanted information, such as smudges, creases, and the noise often present in scans of hand-drawn mechanical drawings” ([Weil, 2017, p. 2](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=2&annotation=H99MXEY2)) #a28ae5
			- - “our first three contributions” ([Weil, 2017, p. 2](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=2&annotation=23LDG3HD)) #a28ae5
			- - “rule-based and heuristic systems” ([Weil, 2017, p. 2](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=2&annotation=PTZT24NV)) #a28ae5
			- - “advancements in deep learning have shown promise in this field” ([Weil, 2017, p. 2](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=2&annotation=NJTKC7HL)) #ffd400
			- *Are this referred to "modern techniques" previously mentioned?*
			- - “The focus of our research is to digitize mechanical drawings with the goal of developing a modular interpretation system that can easily incorporate new annotation standards.” ([Weil, 2017, p. 2](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=2&annotation=8FXPDAYD)) #e56eee
			- - “This refinement reduces the number of shapes to describe the drawing.” ([Weil, 2017, p. 2](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=2&annotation=2LPC3RV4)) #5fb236
			- - “Graph matching involves associating one graph with another” ([Weil, 2017, p. 3](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=3&annotation=39KLPWT9)) #a28ae5
			- - “exact methods” ([Weil, 2017, p. 3](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=3&annotation=9M2LTPVU)) #2ea8e5
			- - “approximate methods, that allow for a certain degree of tolerance [17]” ([Weil, 2017, p. 3](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=3&annotation=EJUIM6W8)) #2ea8e5
			- - “We are particularly interested in the approximate methods as they are easily adaptable and do not require training.” ([Weil, 2017, p. 3](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=3&annotation=6C6TWFRC)) #a28ae5
			- - “This approach involves defining a set of editing rules to transform one graph into another.” ([Weil, 2017, p. 3](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=3&annotation=NLP6B8S9)) #5fb236
			- - “this method calculates a distance between two graphs based on the total cost of actions required for the transformation” ([Weil, 2017, p. 3](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=3&annotation=TC5RCNBI)) #5fb236
			- - “This tracking provides a more detailed understanding of how one graph is transformed into another, which can be particularly useful for explainability.” ([Weil, 2017, p. 3](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=3&annotation=U9X3NY6F)) #ffd400
			- *What is the complexity of this operation?*
			- - “querying of large databases of such drawings.” ([Weil, 2017, p. 3](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=3&annotation=6FV8P74G)) #a28ae5
			- - “This method does not enable precise comparison of information nor effectively highlights the differences between the drawings.” ([Weil, 2017, p. 3](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=3&annotation=E4RYU52W)) #ffd400
			- - “Piping and Instrumentation Diagrams (P&IDs).” ([Weil, 2017, p. 3](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=3&annotation=UQNUENG2)) #5fb236
			- - “Binarized hand-drawn drawing Image” ([Weil, 2017, p. 3](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=3&annotation=K8E4D8DD)) #ffd400
			- *This is the main issue that I see. How can we be sure that hand-drawn images are correctly vectorized? This is the main shortcomings.*
			- - “Metamodel” ([Weil, 2017, p. 3](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=3&annotation=PSLVZFHZ)) #ffd400
			- *What's the use of this?*
			- - “annotations are represented geometrically, they should be regarded as metadata.” ([Weil, 2017, p. 3](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=3&annotation=KN6J3KVE)) #ffd400
			- - “The vectorisation algorithm is considered a parameter of our approach, meaning the detected shapes depend on the specific algorithm used.” ([Weil, 2017, p. 4](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=4&annotation=LBQDSFLQ)) #a28ae5
			- - “we apply a vectorisation algorithm to raster images, resulting in a list of detected shapes.” ([Weil, 2017, p. 4](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=4&annotation=3FCZVD36)) #5fb236
			- - “treating each intersection point as vertices in the graph” ([Weil, 2017, p. 4](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=4&annotation=M7CTH9M3)) #5fb236
			- - “segments s1 and s2 and the intersection i1 are grouped into a hyperedge a1.” ([Weil, 2017, p. 4](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=4&annotation=TT56ENIL)) #5fb236
			- - “4.3 Comparison” ([Weil, 2017, p. 5](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=5&annotation=MDT7BZPX)) #ffd400
			- *All the model comparison work done in the modeling community has been neglected. In the end, once CAD artifacts are represented as graphs, they are models that can be compared by using existing technologhies, that should be considere din the work.*
			- - “function f that maps a node in the initial graph to a corresponding node in the target graph.” ([Weil, 2017, p. 5](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=5&annotation=PCAQ67ZC)) #5fb236
			- - “We will assess whether our comparison method can detect all the dissimilarities that engineers can identify in order to evaluate the comparison approach. We will compare the efficiency of our method with the engineers’ performance to evaluate if our method enhances the overall detection process.” ([Weil, 2017, p. 5](zotero://select/library/items/D8AE4HWY)) ([pdf](zotero://open-pdf/library/items/MDHX56YJ?page=5&annotation=QH9QAZ3X)) #5fb236
	- ### [[Comments]]
		- In the paper, the author introduces a method for comparing hand-drawn mechanical drawings with their corresponding CAD drawings. They suggest automating the verification process to detect differences between the two types of drawings. This method involves represent drawings in terms of graphs, and comparing them to identify inconsistencies.
		- Value: The problem addressed in this paper is significant and highly relevant to industries relying on historical mechanical drawings. This research is driven by real-world needs, thereby providing a valuable contribution to the field.
		- Impact:  By reducing the manual effort required for verification, this approach could lead to substantial time and cost savings. However, further than the practical implications of the approach, the author should emphasize the potential research impact that this work could achieve, e.g., in the modeling field.
		- Originality: In the paper, a new method is proposed that combines different techniques like graph-based structures and GNNs to segment and interpret mechanical drawings. Although the paper shows a good grasp of the relevant literature, encompassing both traditional and deep learning methods, it would be valuable to provide a more detailed comparison with the latest techniques in model comparison, especially within the modeling community. Specifically, when hand-drawn and CAD drawings are represented as graphs, these models can be compared using existing technologies, which should be taken into consideration in the research.
		- Rationale: The proposed approach is well-founded. The authors effectively communicate the importance of accurately transforming mechanical drawings and highlight the shortcomings of current manual verification methods. The decision to use graphs to depict geometric information and to apply graph-matching techniques is well-supported.
		- Research Plan: The research plan is realistic and well-structured. The authors have outlined a clear methodology for graph construction, interpretation, and comparison. They have also proposed a comprehensive validation plan using both CAD drawings and industrial data. However, providing more explicit details on evaluation metrics would enhance the robustness of the plan.
	- ### [[REVIEWS/Notes]]
	- ### YELLOW CONCERNS
	  background-color:: yellow
		- {{query (and [[ffd400]] [[DocSymMod24_9]] )}}
		  collapsed:: true
	- ### ❓️Questions
		- {{query (and [[question]] [[DocSymMod24_9]] )[[question]]}}
		  query-table:: true
		  query-properties:: [:block]