A Method to Enable Automatic Extraction of Cost and Quantity Data from Hierarchical Construction Information Documents to Enable Rapid Digital Comparison and Analysis

Round 1

Reviewer 1 Report

The study is an interesting one. I have only a few concerns.

1. The major contribution of the paper may be described in the introduction section instead of using highlights.

2. Research gaps may be elaborated more.

3. Research questions may be framed and answered.

4. Is the dataset balanced?

5. Can the experiments reproducible?

6. What are the most important features?

7. The limitation of the study should be highlighted.

8. Future research directions may be described.

Author Response

On behalf of the authors of the paper, I wanted to thank you in advance for your time and for the positive review received. As it has been requested, the following issues have been reviewed:

1. The major contribution of the paper may be described in the introduction section instead of using highlights.

The highlights of the article have been excluded and the main strengths of the approach have been incluede in the introduction and in the novelty section.

2. Research gaps may be elaborated more.

The reviewed literature and the introduction section have been reviewed to ensure that the research gaps are being elaborated and the problem faced is well stated.

3. Is the dataset balanced?

That is indeed a good concern. The data comes from two different projects where the costs categories have been irregularly distributed among the number of instances. However, the big size of two datasets and the preprocessing step are able to cope with the balance problem. This have been explained in detail in section 2.2 and 2.3.

4. Can the experiments be reproducible?

That is a very important concern. Section 2.1 has been reviewed to ensure that all the input data have been describe with enough detail allowing for its replicability. Additionally the different steps encompassing the methodology have also been reviewed to ensure that they are descriptive enough.

5. What are the most important features?

The most important features of the method that are worth highlighting would be Firstly, the completeness of the solution being able to extract, preprocess data, apply NLP and being able to classify the resulting information. Secondly, its strong validation since the solution has been tested on two different data sets. Finally, the novelty of its scope being able to bring a novel solution with a combination of different technologies assembled in a very specific manner.

6. The limitation of the study should be highlighted.

Thank you very much for your feedback. The limitations of the study have been included in a new section 2.3 to present the whole picture to the reader expressing not only the benefits but also the limitations of the suggested method.

7. Future research directions may be described.

To cope with this issue.  A new section 4.1 has been including specifying the future research work that could be implemented based on the main findings reported by the paper.

Reviewer 2 Report

Abstract

The article describes a novel method for the automatic extraction and organization of construction cost and quantity data. This method enables rapid digital comparison and analysis, overcoming inconsistencies, misclassifications, and time-consuming manual processes. It involves a two-step process of data mining followed by classification and processing into a common format, with a reported success rate of 97.5%.

Introduction

The introduction provides a historical context for the issue, emphasizing the lack of standardization in the construction industry and the shift in structuring information from trade-based to element-based. It acknowledges the challenges of digital transfer and manual data classification. It also presents the need for automated data classification and previews the end-to-end methodology of the article. In addition, the introduction contains a section on related work that covers existing methods, criticisms, and the novelty of the presented approach.

Method

The article’s main body lays out the two-step method in detail:

1. Data Mining: This first step involves reviewing the input document to determine its structure, based on various parameters such as position, format, sequence, and content of the data.
2. Classification and Processing: The second step involves processing and classifying the extracted data using data science and expert knowledge to fit a common format.

The approach is flexible enough to handle different file types and structures and aims to automate the entire process, allowing direct comparisons across projects.

Related Work

This section summarizes the literature and previous research related to the article's topic, including studies that have applied different techniques such as Natural Language Processing (NLP), decision trees, and support vector machines to similar problems. It highlights both the successes and limitations of previous approaches, establishing a context for the novel method presented in the article.

Novelty of the Method

The article's method is touted as innovative due to its end-to-end approach, strong validation, and unique combination of existing technologies. These features are explained as distinct from previous attempts to tackle similar problems.

Conclusion

The conclusion reiterates the main contributions of the paper, specifically the creation of an automated, accurate method for data extraction and classification that emulates expert knowledge without human intervention. The success rate of 97.5% in a real-case scenario attests to the method’s robustness, with the remaining 2.5% of inaccuracies attributed to input file irregularities.

Overall Analysis

The article successfully introduces a method to resolve a significant challenge within the construction industry. It builds on previous work while adding significant innovation in terms of automation and accuracy. The clarity and structure of the article facilitate understanding, and the method's practical application is well-supported by empirical data. Potential areas for improvement might include a more detailed exploration of the limitations or potential biases of the method and how it might be adapted or extended to other contexts. The impact of the method on industry practices and how it might interface with existing systems and processes could also be explored in more depth. Overall, the article makes a valuable contribution to the field, offering a tangible solution to a long-standing problem.

Abstract

The article describes a novel method for the automatic extraction and organization of construction cost and quantity data. This method enables rapid digital comparison and analysis, overcoming inconsistencies, misclassifications, and time-consuming manual processes. It involves a two-step process of data mining followed by classification and processing into a common format, with a reported success rate of 97.5%.

Introduction

The introduction provides a historical context for the issue, emphasizing the lack of standardization in the construction industry and the shift in structuring information from trade-based to element-based. It acknowledges the challenges of digital transfer and manual data classification. It also presents the need for automated data classification and previews the end-to-end methodology of the article. In addition, the introduction contains a section on related work that covers existing methods, criticisms, and the novelty of the presented approach.

Method

The article’s main body lays out the two-step method in detail:

1. Data Mining: This first step involves reviewing the input document to determine its structure, based on various parameters such as position, format, sequence, and content of the data.
2. Classification and Processing: The second step involves processing and classifying the extracted data using data science and expert knowledge to fit a common format.

The approach is flexible enough to handle different file types and structures and aims to automate the entire process, allowing direct comparisons across projects.

Related Work

This section summarizes the literature and previous research related to the article's topic, including studies that have applied different techniques such as Natural Language Processing (NLP), decision trees, and support vector machines to similar problems. It highlights both the successes and limitations of previous approaches, establishing a context for the novel method presented in the article.

Novelty of the Method

The article's method is touted as innovative due to its end-to-end approach, strong validation, and unique combination of existing technologies. These features are explained as distinct from previous attempts to tackle similar problems.

Conclusion

The conclusion reiterates the main contributions of the paper, specifically the creation of an automated, accurate method for data extraction and classification that emulates expert knowledge without human intervention. The success rate of 97.5% in a real-case scenario attests to the method’s robustness, with the remaining 2.5% of inaccuracies attributed to input file irregularities.

Overall Analysis

The article successfully introduces a method to resolve a significant challenge within the construction industry. It builds on previous work while adding significant innovation in terms of automation and accuracy. The clarity and structure of the article facilitate understanding, and the method's practical application is well-supported by empirical data. Potential areas for improvement might include a more detailed exploration of the limitations or potential biases of the method and how it might be adapted or extended to other contexts. The impact of the method on industry practices and how it might interface with existing systems and processes could also be explored in more depth. Overall, the article makes a valuable contribution to the field, offering a tangible solution to a long-standing problem.

Author Response

On behalf of all the authors, I wanted to express our gratitude for your time and for the positive feedback received.

As requested. two new sections have been included to express the main limitations of the study. The main goal would be to express the whole picture and not only its strengths and its novelty based on the reviewed literature. Secondly, another section has been included to express future work directions based on the main findings of the study. Finally, the impact and the benefits of the suggested method have been stressed on the remaining sections of the paper.