A method for intelligent auditing and detecting quality of architectural scheme text

Abstract

The application discloses a kind of building scheme text quality intelligent auditing detection method, belong to building auditing technical field, including collection integration historical scheme, generating auditing details, reading to be audited building scheme, auditing building scheme, feedback modification scheme and artificial review.The application generates a duplicate checking table by extracting the atlas, main content and keywords in the historical building scheme, checks the duplicate for the building scheme to be audited, avoids plagiarism, ensures the quality of the building scheme to be audited, and at the same time, if the duplicate checking rate of the same building scheme exceeds three times, the building scheme is prohibited to be audited within one week, and the system is uploaded for record, indirectly reduces the probability of plagiarism scheme, the data of excellent construction scheme in the historical database is used to train the text similarity algorithm model, the writing quality of the building scheme text is audited and judged, and the algorithm model is used to audit the building scheme, without manual auditing by staff, improving the auditing efficiency.

Description

Technical Field

[0001] This invention relates to the field of architectural review technology, and in particular to an intelligent review and detection method for the quality of architectural design documents. Background Technology

[0002] Construction plans are the implementation schemes for engineering projects. They are a large number of documents written by the technical personnel of construction companies on a daily basis, including organizational structure plans, personnel composition plans, technical plans, safety plans, material supply plans, calculation sheets, etc.

[0003] Current methods still rely on manual review and approval of project construction plans. This manual review requires reading the plan word by word, understanding the plan and its diagrams, and making corresponding annotations and modifications. Moreover, there may be multiple architectural plans, which makes the workload extremely heavy for staff. This is not only inefficient for project construction but also lacks accuracy. Therefore, an intelligent review and detection method for architectural plan text quality is proposed to solve the above problems. Summary of the Invention

[0004] The purpose of this invention is to solve the problems in the prior art by proposing an intelligent review and detection method for the quality of architectural design texts.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A method for intelligent review and detection of architectural design document quality includes the following steps:

[0007] S1. Collection and Integration of Historical Schemes: Historical building schemes are collected, and then the scheme number, relevant authors and publication time are edited. Finally, the schemes are integrated and classified according to different types of construction schemes. At the same time, the diagrams, main contents and keywords in the historical building schemes are extracted to generate a duplicate check table.

[0008] S2. Generate review details: By using data from historical databases of approved and excellent construction plans, a text similarity algorithm model is trained to judge the writing quality of architectural plan texts. At the same time, various problems and solutions that have appeared in historical plans are collected and compiled into a revision table. Then, a rating table is generated based on the pass rate in the algorithm model.

[0009] S3. Read the architectural plan to be reviewed: Use a scanner to scan and read the architectural plan to be reviewed, convert the text format of the architectural plan, then extract the diagrams, main content and keywords from the architectural plan to be reviewed, and classify the architectural plan to be reviewed. Compare the extracted diagrams, main content and keywords with the diagrams, main content and keywords of the same type of historical plan in the duplication table in S1 one by one to realize the duplication function of the architectural plan to be reviewed.

[0010] S4. Reviewing architectural plans: The team developed a word vector training model for the semantic level of the architectural field, which can review the relevance of the title and content of architectural plans. After the review is passed, the graph, main content and keywords of the architectural plan to be reviewed extracted in S3 are compared with the algorithm model in S1. During the review process, paragraphs are marked in the content of the plan, the unqualified parts of the content are marked and the judgment basis is stated. At the same time, the number of the standard is calculated to rate the architectural plan.

[0011] S5. Feedback on Modification Scheme: Based on the revision table generated in S1, provide corresponding modification suggestions at the unqualified marks in S4, and generate an audit form for user reference;

[0012] S6. Manual Review: Provide the review forms from S5 to professional reviewers for review. By combining the results of intelligent review and manual review, the accuracy of the architectural plan review can be ensured.

[0013] Preferably, the integration method of historical schemes in S1 is that each architectural scheme corresponds one-to-one with the corresponding author and publication time, and the main contents of the extracted historical architectural schemes include the basis for compilation, project overview, architectural design overview, calculation sheets, etc.

[0014] Preferably, the rating criteria in S2 are as follows: ≥90% compliance rate is Grade A, ≥80% compliance rate is Grade B, ≥70% compliance rate is Grade C, ≥60% compliance rate is Grade D, and ≤60% compliance rate is unqualified.

[0015] Preferably, if the same building scheme fails the duplication check more than three times in S3, the building scheme will be prohibited from being reviewed for one week and will be uploaded to the system for record-keeping. The prohibition period will increase week by week as the number of failures increases.

[0016] Preferably, in step S3, for architectural plans that pass the plagiarism check, the system will provide feedback on the review progress, indicating whether the review is pending, in progress, or completed.

[0017] Preferably, in step S3, converting the architectural plan text format involves converting PDF, TXT, or other formats into Word document format, generating an approval number, and integrating it with the architectural plan for archiving, facilitating subsequent retrieval of the approved plan.

[0018] Preferably, the content of the manual review in S6 mainly includes the compliance of the compilation basis, the completeness of the project overview, the standardization of the architectural design parameters, and the accuracy of the calculation results.

[0019] Preferably, a plagiarism rate of ≤30% indicates that the architectural design is not suspected of plagiarism; otherwise, it indicates that the architectural design is suspected of plagiarism, and the design will be rejected and the design number, design name, publication time, and author of the historical design with the highest overlap rate will be provided as feedback.

[0020] Preferably, in step S6, if there is a discrepancy between the opinions expressed by the manual review and the intelligent review, the discrepancy in the solution is transferred to a third-party testing agency for re-review, and the result of the third-party testing agency shall prevail.

[0021] Preferably, if a proposed solution fails the review process more than three times in step S4, the solution will be restricted from being submitted again for three days.

[0022] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0023] 1. This method generates a plagiarism check form by extracting the drawings, main content, and keywords from historical architectural plans. This form is used to check for plagiarism in plans to be reviewed, thereby preventing plagiarism and ensuring the quality of the architectural plans to be reviewed. If the plagiarism rate of the same architectural plan exceeds three times, the plan will be prohibited from being reviewed for one week and will be uploaded to the system for record-keeping, indirectly reducing the probability of plagiarism.

[0024] 2. This solution trains a text similarity algorithm model using data from historical databases of approved and excellent construction plans. This model is used to assess the writing quality of architectural plans, eliminating the need for manual review and improving efficiency.

[0025] 3. This plan, by marking sections of the plan during the review process and indicating the criteria for judging non-compliance, facilitates users' subsequent understanding and modification, helps users quickly improve their architectural plans, and rates the reviewed architectural plans to give users a clear understanding of the plans.

[0026] 4. This invention ensures the accuracy of architectural plan review by setting up manual review and combining the results of intelligent review and manual review. Attached Figure Description

[0027] Figure 1 This is a system diagram of an intelligent review and detection method for architectural design text quality proposed in this invention;

[0028] Figure 2 This is a flowchart illustrating the intelligent review and detection method for architectural design text quality proposed in this invention. Detailed Implementation

[0029] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0030] Reference Figure 1-2 A method for intelligent review and detection of architectural design text quality includes the following steps:

[0031] Historical plans are collected and integrated: historical building plans are collected, and then the plan number, relevant authors and publication time are edited. Finally, the plans are integrated and classified according to different types of construction plans. At the same time, the diagrams, main contents and keywords in the historical building plans are extracted to generate a duplicate check table.

[0032] The historical plans were integrated by matching each architectural plan with its corresponding author and publication date. The main contents extracted from the historical architectural plans include the basis for compilation, project overview, architectural design overview, calculation sheets, etc.

[0033] It should be noted that a plagiarism check form is generated by extracting the drawings, main content, and keywords from historical architectural plans. This form is used to check for plagiarism in the plans to be reviewed, thus preventing plagiarism and ensuring the quality of the architectural plans to be reviewed. If the plagiarism rate of the same architectural plan exceeds three times, the architectural plan will be prohibited from being reviewed for one week, and the information will be uploaded to the system for record-keeping, indirectly reducing the probability of plagiarism.

[0034] Generate review details: By using data from historical databases of approved and excellent construction plans, a text similarity algorithm model is trained to judge the writing quality of architectural plan texts. At the same time, various problems and solutions that have appeared in historical plans are collected and compiled into a revision table. Then, a rating table is generated based on the pass rate in the algorithm model.

[0035] The rating criteria are as follows: ≥90% compliance rate is Grade A, ≥80% compliance rate is Grade B, ≥70% compliance rate is Grade C, ≥60% compliance rate is Grade D, and ≤60% compliance rate is unqualified.

[0036] It should be noted that the text similarity algorithm model is trained by using data from historical databases of approved and excellent construction plans to assess the writing quality of architectural plans. The algorithm model is used to review architectural plans, thus eliminating the need for manual review by staff and improving review efficiency.

[0037] Read the architectural design to be reviewed: Use a scanner to scan and read the architectural design to be reviewed, convert the text format of the architectural design, and then extract the diagrams, main content and keywords from the architectural design to be reviewed. The architectural design to be reviewed is classified and the extracted diagrams, main content and keywords are compared with the diagrams, main content and keywords of the same type of historical design in the plagiarism check table one by one. If the plagiarism rate is ≤30%, it means that the architectural design is not suspected of plagiarism. Otherwise, it means that the architectural design is suspected of plagiarism, and the review is rejected. The design number, design name, publication time and author of the historical design with the highest overlap rate are provided as feedback.

[0038] If the same architectural design fails the plagiarism check more than three times, the design will be prohibited from being reviewed for one week and will be uploaded to the system for record-keeping. The prohibition period will increase week by week as the number of failures increases.

[0039] It should be noted that by extracting the diagrams, main content, and keywords from the architectural design to be reviewed and comparing them one by one with the diagrams, main content, and keywords of the comparison file, the efficiency of plagiarism detection can be greatly improved, thereby avoiding plagiarism and ensuring the quality of the architectural design to be reviewed. At the same time, setting up penalty measures will make users who plagiarize feel in awe, thereby indirectly reducing the probability of plagiarized designs appearing.

[0040] Architectural Design Review: The team developed a semantic-level word vector training model for the architectural field, which can review the relevance of the title and content of architectural design proposals. After approval, the extracted graphs, main content, and keywords of the architectural design proposal to be reviewed are compared with the algorithm model. During the review process, paragraphs in the proposal are annotated, highlighting any unqualified parts and stating the basis for the judgment. At the same time, the number of times the standard is met is calculated to rate the architectural design proposal. Feedback and Modification: Based on the generated revision table, corresponding modification suggestions are given at the marked unqualified points, and an review table is generated for user reference.

[0041] It should be noted that by annotating the sections of the design proposal during the review process and indicating the basis for judging the unsatisfactory parts, it is easier for users to understand and modify the proposals later, helping them to quickly improve their architectural designs. At the same time, the reviewed architectural designs are rated, which allows users to have an intuitive understanding of the architectural designs.

[0042] Manual review: The review form is provided to professional reviewers for review. By combining the results of intelligent review and manual review, the accuracy of the architectural plan review is ensured. The content of manual review mainly includes the compliance of the basis for preparation, the completeness of the project overview, the standardization of architectural design parameters, and the accuracy of the calculation results.

[0043] It should be noted that manual review can serve as a verification tool for intelligent review, and combining the results of intelligent review and manual review can produce a more objective review report, thereby improving the accuracy of the review.

[0044] For architectural designs that pass the plagiarism check, the system will provide feedback on the review progress, indicating whether the design is awaiting review, under review, or completed, allowing users to check the review progress at any time.

[0045] Converting architectural design plans into text format involves converting PDF, TXT, and other formats into Word document format, generating an approval number, and integrating it with the architectural design plan for archiving, facilitating subsequent retrieval of the approved plan.

[0046] When discrepancies arise between manual and intelligent reviews, the areas in the plan that contain these discrepancies are transferred to a third-party testing agency for re-review. The results from the third-party testing agency will be considered final to ensure the accuracy of the review.

[0047] If a proposal fails the review more than 3 times, it will be restricted from being submitted again for 3 days to encourage users to revise their proposals carefully.

[0048] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for intelligent review and detection of architectural design text quality, characterized in that, Includes the following steps: S1. Collection and Integration of Historical Schemes: Historical building schemes are collected, and then the scheme number, relevant authors and publication time are edited. Finally, the schemes are integrated and classified according to different types of construction schemes. At the same time, the diagrams, main contents and keywords in the historical building schemes are extracted to generate a duplicate check table. S2. Generate review details: By using data from historical databases of approved and excellent construction plans, a text similarity algorithm model is trained to judge the writing quality of architectural plan texts. At the same time, various problems and solutions that have appeared in historical plans are collected and compiled into a revision table. Then, a rating table is generated based on the pass rate in the algorithm model. S3. Read the architectural plan to be reviewed: Use a scanner to scan and read the architectural plan to be reviewed, convert the text format of the architectural plan, then extract the diagrams, main content and keywords from the architectural plan to be reviewed, and classify the architectural plan to be reviewed. Compare the extracted diagrams, main content and keywords with the diagrams, main content and keywords of the same type of historical plan in the duplication table in S1 one by one to realize the duplication function of the architectural plan to be reviewed. S4. Reviewing architectural plans: The team developed a word vector training model for the semantic level of the architectural field, which can review the relevance of the title and content of architectural plans. After the review is passed, the graph, main content and keywords of the architectural plan to be reviewed extracted in S3 are compared with the algorithm model in S2. During the review process, paragraphs are marked in the content of the plan, the unqualified parts of the content are marked and the judgment basis is stated. At the same time, the number of the standard is calculated to rate the architectural plan. S5. Feedback on Modification Scheme: Based on the revision table generated in S1, provide corresponding modification suggestions at the unqualified marks in S4, and generate an audit form for user reference; S6. Manual Review: Provide the review forms from S5 to professional reviewers for review. By combining the results of intelligent review and manual review, the accuracy of the architectural plan review can be ensured.

2. The intelligent review and detection method for architectural design text quality according to claim 1, characterized in that: The integration method of historical schemes in S1 is to match each architectural scheme with its corresponding author and publication time. The main contents of the extracted historical architectural schemes include the basis for compilation, project overview, architectural design overview, and calculation sheets.

3. The intelligent review and detection method for architectural design text quality according to claim 1, characterized in that: The rating criteria in S2 are as follows: ≥90% compliance rate is Grade A, ≥80% compliance rate is Grade B, ≥70% compliance rate is Grade C, ≥60% compliance rate is Grade D, and ≤60% compliance rate is unqualified.

4. The intelligent review and detection method for architectural design text quality according to claim 1, characterized in that: If the same building design fails the duplication check more than three times in S3, the building design will be prohibited from being reviewed for one week and will be uploaded to the system for record-keeping. The prohibition period will increase week by week as the number of failures increases.

5. The intelligent review and detection method for architectural design text quality according to claim 1, characterized in that: In S3, for architectural plans that pass the plagiarism check, the system will provide feedback on the review progress, indicating whether the review is pending, in progress, or completed.

6. The intelligent review and detection method for architectural design text quality according to claim 1, characterized in that: The S3 step involves converting the architectural plan text format from PDF, TXT, and other formats to Word document format, and generating an approval number to integrate and archive the architectural plan for easy retrieval later.

7. The intelligent review and detection method for architectural design text quality according to claim 1, characterized in that: The content of manual review in S6 mainly includes the compliance of the basis for preparation, the completeness of the project overview, the standardization of architectural design parameters, and the accuracy of the calculation results.

8. The intelligent review and detection method for architectural design text quality according to claim 4, characterized in that: A plagiarism rate of ≤30% indicates that the architectural design is not suspected of plagiarism; otherwise, it indicates that the architectural design is suspected of plagiarism, and the design will be rejected and the design number, name, publication time, and author of the historical design with the highest overlap rate will be provided as feedback.

9. The intelligent review and detection method for architectural design text quality according to claim 1, characterized in that: In step S6, if there is a discrepancy between the opinions expressed in the manual review and the intelligent review, the discrepancy in the solution will be transferred to a third-party testing agency for re-review, and the results of the third-party testing agency shall prevail.

10. The intelligent review and detection method for architectural design text quality according to claim 1, characterized in that: If a proposed solution fails the review process more than three times in S4, the solution will be restricted from being submitted again for three days.

Images