Structured natural language suitable review method, apparatus and electronic device

Abstract

The application relates to a structured natural language applicable review method and device and electronic equipment, comprising: replacing a second clause of each structured natural language sentence in an original review result report file with a third clause meeting a preset judgment conversion condition, splicing a first clause and the third clause of each structured natural language sentence to obtain a spliced structured natural language sentence, performing an applicable review operation on each spliced structured natural language sentence based on a mapping relationship between each structured natural language sentence and each spliced structured natural language sentence and a preset compliance judgment formula to obtain an applicable review result of each structured natural language sentence. Thus, the review method in the related art can cause deviation of the compliance judgment result, cannot accurately locate the design hidden compliance risk, and is not conducive to the staff to quickly complete problem rectification and specification verification, and the accuracy of the building design compliance review can be significantly improved.

Description

Technical Field

[0001] This application relates to the field of automated compliance review technology for construction projects, and in particular to a structured natural language-based review method, apparatus, and electronic device. Background Technology

[0002] Currently, automated compliance review technology is developing rapidly and has been widely applied to architectural design compliance review. Its essence is to use computer computing power and storage capacity, combined with algorithms such as knowledge graphs and compiler principles, to automatically verify whether design schemes in specific fields meet the requirements of corresponding regulatory provisions in the construction industry.

[0003] Among the related technologies, the architectural design compliance review is mainly based on Structured Natural Language (SNL). The SNL review logic it adopts is a typical binary logic: a single SNL rule can only output two types of review results: pass or fail.

[0004] However, when the relevant technology performs architectural design compliance reviews based on SNL, it cannot distinguish the reasons for the inapplicable review results. That is, it cannot distinguish whether the building is not within the scope of the review rules or whether the architectural design does not meet the review standards, resulting in biased compliance judgments. It cannot accurately locate hidden compliance risks in the design, which is not conducive to staff quickly completing problem rectification and standard verification. This needs to be resolved urgently. Summary of the Invention

[0005] This application provides a structured natural language review method, apparatus, and electronic device to address the problems in related technologies where review methods are prone to biased compliance judgments, fail to accurately locate hidden compliance risks in designs, and hinder staff from quickly completing problem rectification and standard verification. This application can significantly improve the accuracy of architectural design compliance review.

[0006] The first aspect of this application provides an applicable review method for structured natural language processing, comprising the following steps: In response to compliance review requirements, obtain the original review result report documents corresponding to the compliance review requirements, and obtain a set of structured natural language statements based on the original review result report documents; Determine the first and second clauses of each structured natural language statement in the set of structured natural language statements, and replace the second clause of each structured natural language statement with a third clause that satisfies the preset judgment conversion conditions. Then, concatenate the first and third clauses of each structured natural language statement to obtain the corresponding concatenated structured natural language statement. The mapping relationship between each structured natural language statement and each concatenated structured natural language statement is determined. Based on the mapping relationship and the preset compliance judgment formula, the applicable review operation is performed on each concatenated structured natural language statement to obtain the applicable review result of each structured natural language statement.

[0007] Optionally, in some embodiments, a set of structured natural language statements is obtained based on the original review result report document, including: Read the JSON string from the original review result report file; The JSON string is parsed to obtain the review result report object, and all review task result objects stored in the review result report object are traversed to obtain the structured natural language statement stored in each review task result object; A set of structured natural language statements is generated based on the structured natural language statements stored in the result object of each review task.

[0008] Optionally, in some embodiments, the second clause of each structured natural language statement is replaced with a third clause that satisfies a preset conversion condition, including: Identify the first subject appearing in the first clause of each structured natural language statement; Based on the first occurrence of the subject in the first clause of each structured natural language statement, the second clause of the corresponding structured natural language statement is replaced with the third clause that satisfies the preset judgment conversion conditions.

[0009] Optionally, in some embodiments, based on the mapping relationship and a preset compliance determination formula, an applicability review operation is performed on each concatenated structured natural language statement to obtain the applicability review result of each structured natural language statement, including: Based on each concatenated structured natural language statement, the first structured natural language statement containing preset computational rules and the second structured natural language statement not containing preset computational rules are selected. Based on the preset compliance judgment formula, the first structured natural language statement is reviewed using the preset SPARQL SELECT statement to obtain the first review result, and the second structured natural language statement is reviewed using the preset SPARQL ASK statement to obtain the second review result. The applicable review results for each structured natural language statement are obtained based on the first and second review results.

[0010] Optionally, in some embodiments, the preset compliance determination formula is: ; in, The first to nth components in the first clause; n represents the first to the mth components in the second clause; n and m are both positive integers.

[0011] Optionally, in some embodiments, after performing an applicability review operation on each concatenated structured natural language statement based on the mapping relationship and a preset compliance determination formula to obtain the applicability review result of each structured natural language statement, the method further includes: Based on the applicable review results of each structured natural language statement, target structured natural language statements that are not applicable to review are selected. The review results corresponding to the target structured natural language statements are stored in the corresponding fields of the original review result report file.

[0012] A second aspect of this application provides an applicability review device for structured natural language processing, comprising: The acquisition module is used to respond to compliance review requirements by acquiring the original review result report file corresponding to the compliance review requirements, and obtaining a set of structured natural language statements based on the original review result report file; The replacement module is used to determine the first clause and the second clause of each structured natural language statement in the set of structured natural language statements, and replace the second clause of each structured natural language statement with the third clause that satisfies the preset judgment conversion conditions, and concatenate the first clause and the third clause of each structured natural language statement to obtain the corresponding concatenated structured natural language statement. The review module is used to determine the mapping relationship between each structured natural language statement and each concatenated structured natural language statement, and based on the mapping relationship and the preset compliance judgment formula, to perform applicable review operations on each concatenated structured natural language statement to obtain the applicable review result of each structured natural language statement.

[0013] Optionally, in some embodiments, the acquisition module is specifically used for: Read the JSON string from the original review result report file; The JSON string is parsed to obtain the review result report object, and all review task result objects stored in the review result report object are traversed to obtain the structured natural language statement stored in each review task result object; A set of structured natural language statements is generated based on the structured natural language statements stored in the result object of each review task.

[0014] Optionally, in some embodiments, the replacement module is specifically used for: Identify the first subject appearing in the first clause of each structured natural language statement; Based on the first occurrence of the subject in the first clause of each structured natural language statement, the second clause of the corresponding structured natural language statement is replaced with the third clause that satisfies the preset judgment conversion conditions.

[0015] Optionally, in some embodiments, the review module is specifically used for: Based on each concatenated structured natural language statement, the first structured natural language statement containing preset computational rules and the second structured natural language statement not containing preset computational rules are selected. Based on the preset compliance judgment formula, the first structured natural language statement is reviewed using the preset SPARQL SELECT statement to obtain the first review result, and the second structured natural language statement is reviewed using the preset SPARQL ASK statement to obtain the second review result. The applicable review results for each structured natural language statement are obtained based on the first and second review results.

[0016] Optionally, in some embodiments, the preset compliance determination formula is: ; in, The first to nth components in the first clause; n represents the first to the mth components in the second clause; n and m are both positive integers.

[0017] Optionally, in some embodiments, after performing an applicability review operation on each concatenated structured natural language statement based on the mapping relationship and a preset compliance determination formula to obtain the applicability review result for each structured natural language statement, the review module is further configured to: Based on the applicable review results of each structured natural language statement, target structured natural language statements that are not applicable to review are selected. The review results corresponding to the target structured natural language statements are stored in the corresponding fields of the original review result report file.

[0018] A third aspect of this application provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the applicable review method for structured natural language as described in the first aspect embodiment.

[0019] A fourth aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement the applicable review method for structured natural language as described in the first aspect embodiment.

[0020] Therefore, the embodiments of this application have at least the following beneficial effects: (1) This application provides an applicable review method based on SNL clause replacement, which can efficiently and accurately review various types and styles of SNL statements. It can distinguish between truly effective statements in the reviewed SNL and statements that are passed due to inapplicability, providing users with more refined review results. This makes the statistical data of the review results more consistent with the actual situation, avoids the inclusion of inapplicable SNL in the statistics, which leads to an artificially high pass rate, and solves the problem that the SNL review engine in related technologies cannot review the "if" part of the SNL statement separately.

[0021] (2) This application proposes an application review time efficiency optimization method based on SPARQL ASK statement, which can improve the execution time efficiency of SNL application review. At the same time, by adopting the operation logic of filter-like, only the review result report output by the original review engine is further processed, which can minimize the impact on the overall operation process of the original SNL review engine.

[0022] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0023] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 A flowchart illustrating an applicable review method for structured natural language according to an embodiment of this application; Figure 2 A flowchart illustrating an applicable review method for structured natural language according to one embodiment of this application; Figure 3 This is a schematic diagram illustrating a clause replacement according to an embodiment of this application; Figure 4 This is a schematic diagram illustrating the application of examination results to a single structured natural language statement according to an embodiment of this application; Figure 5 This is a schematic diagram of a smart adaptive review report provided according to an embodiment of this application; Figure 6 A flowchart illustrating an applicable review optimization method for structured natural language according to an embodiment of this application; Figure 7 This is a flowchart of an optimized applicable review method based on SPARQL ASK statements according to an embodiment of this application; Figure 8 A block diagram illustrating an applicable review device for structured natural language processing according to an embodiment of this application; Figure 9 This is a block diagram of a structured natural language review device according to an embodiment of this application; Figure 10 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0024] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0025] The following description, with reference to the accompanying drawings, describes an applicable method, apparatus, and electronic device for reviewing structured natural language according to embodiments of this application.

[0026] Before introducing the applicable examination method for structured natural language in this application, let's briefly introduce the relevant content of SNL statements.

[0027] Specifically, SNL (Semantic Language) is a domain-specific language that lies between computer code and natural language. It uses "if...then..." statements to represent architectural clauses in a computer-generated manner, offering both readability and comprehensibility. Conducting architectural design compliance reviews based on SNL primarily involves four processes: converting clauses in the natural language specification into SNL statements that computers can understand and process; extracting architectural semantic information models from BIM models or PDF drawings; running algorithms to review the SNL statements on the semantic model; and generating a final review report by arranging and organizing the review results according to the clauses in the natural language specification. Adopting an automated compliance review model significantly improves review efficiency, avoids errors that may occur during manual review, and to some extent avoids ambiguity caused by unclear natural language clause descriptions, ensuring the reliability and effectiveness of the compliance review results.

[0028] For example, suppose the relevant regulation states that fire protection electrical systems in civil buildings with a height greater than 150 meters should be supplied with electricity at the "Special Grade" load level. This regulation can be expressed using the SNL statement: If the building name includes "civil" and the building height is >150 meters, then the building's fire protection electrical load level is equal to "Special Grade". The application review of this SNL statement involves determining whether the building under review meets the requirement of the SNL if part (i.e., if the building name includes "civil" and the building height is >150 meters). If the building name includes "civil" and the building height is greater than 150 meters, then the SNL statement applies; otherwise, the SNL statement does not apply.

[0029] However, the SNL review logic used by the relevant technology is a typical binary logic: a single SNL rule can only output two review results: pass or fail. It cannot determine whether the review is passed because the SNL is "applicable" and the building meets the specific provisions of the "then" part of the SNL, or because the SNL is "not applicable." In other words, the review technology cannot perform an applicability review of the SNL.

[0030] Based on the aforementioned issues, this application provides an applicable review method for structured natural language (SNR). This method replaces the second clause of each SNR statement in the original review result report with a third clause that meets preset judgment conversion conditions. It then concatenates the first and third clauses of each SNR statement to obtain a corresponding concatenated SNR statement. Based on a preset compliance judgment formula, an applicable review operation is performed on each concatenated SNR statement to obtain its applicable review result. This solves the problems of biased compliance judgment results, inaccurate identification of hidden compliance risks in designs, and difficulties in quickly completing problem rectification and standard verification by staff in related technologies. It significantly improves the accuracy of architectural design compliance review.

[0031] Specifically, Figure 1 A flowchart illustrating the applicable review method for structured natural language provided in this application embodiment.

[0032] like Figure 1 As shown, the applicable review method for structured natural language processing includes the following steps: In step S101, in response to the compliance review requirement, the original review result report file corresponding to the compliance review requirement is obtained, and a set of structured natural language statements is obtained based on the original review result report file.

[0033] In some embodiments, obtaining a set of structured natural language statements based on the original review result report file includes: reading a JSON string from the original review result report file; parsing the JSON string to obtain a review result report object, and traversing all review task result objects stored in the review result report object to obtain the structured natural language statements stored in each review task result object; and generating a set of structured natural language statements based on the structured natural language statements stored in each review task result object.

[0034] Specifically, this embodiment first requires reading and parsing the original examination result report file. After reading the JSON string from the original examination result report file into memory, a JSON parsing library is called to parse the JSON string to obtain the examination result report object. Then, all examination task result objects stored in the examination result report object are traversed, and the structured natural language statements stored in each examination task result object are extracted. These structured natural language statements are then collected and stored in a set of structured natural language statements in memory.

[0035] In step S102, the first clause and the second clause of each structured natural language statement in the set of structured natural language statements are determined, and the second clause of each structured natural language statement is replaced with a third clause that satisfies the preset judgment conversion condition. The corresponding concatenated structured natural language statement is obtained by splicing the first clause and the third clause of each structured natural language statement.

[0036] In some embodiments, the second clause of each structured natural language statement is replaced with a third clause that satisfies a preset conversion condition. This includes: determining the first subject appearing in the first clause of each structured natural language statement; and replacing the second clause of the corresponding structured natural language statement with a third clause that satisfies the preset conversion condition based on the first subject appearing in the first clause of each structured natural language statement.

[0037] The first clause is the conditional clause between "if" and "then" in each structured natural language statement; the second clause is the conditional clause after "then" in each structured natural language statement; the third clause is the new statement obtained after replacing the second clause; the preset judgment conversion conditions can be conditions pre-set by the user, conditions obtained through a limited number of experiments, or conditions obtained through a limited number of computer simulations, and are not specifically limited here.

[0038] Specifically, this embodiment of the application requires performing clause replacement operations on structured natural language statements, while simultaneously saving the mapping relationship between the concatenated structured natural language statements (i.e., the SNL statements after clause replacement) and the original structured natural language statements (i.e., the original SNL statements). The set of structured natural language statements in memory is traversed, the first subject appearing in each structured natural language statement is found, and a third clause (i.e., a clause that is definitely wrong) is generated based on this subject. The first clause (i.e., the clause of the if part) and the third clause of the structured natural language statement are then concatenated to obtain the concatenated structured natural language statement. Furthermore, all concatenated structured natural language statements are written into the clause replacement SNL file. Simultaneously, the mapping relationship between each structured natural language statement and its corresponding concatenated structured natural language statement is saved in memory.

[0039] In step S103, the mapping relationship between each structured natural language statement and each concatenated structured natural language statement is determined. Based on the mapping relationship and the preset compliance judgment formula, the applicable review operation is performed on each concatenated structured natural language statement to obtain the applicable review result of each structured natural language statement.

[0040] In some embodiments, based on mapping relationships and preset compliance determination formulas, an applicability review operation is performed on each concatenated structured natural language statement to obtain an applicability review result for each structured natural language statement. This includes: based on each concatenated structured natural language statement, selecting a first structured natural language statement containing preset computational rules and a second structured natural language statement not containing preset computational rules; based on the preset compliance determination formula, reviewing the first structured natural language statement using a preset SPARQL SELECT statement to obtain a first review result, and reviewing the second structured natural language statement using a preset SPARQL ASK statement to obtain a second review result; and obtaining an applicability review result for each structured natural language statement based on the first and second review results.

[0041] Specifically, this application embodiment requires the use of an examination engine to perform an applicability review on the concatenated structured natural language statements. The specified building model file and clause replacement SNL file are read, and based on a preset compliance determination formula, an applicability review is performed on each concatenated structured natural language statement in the building model, obtaining a pass or fail applicability review result for each structured natural language statement. The preset compliance determination formula is as follows: ; in, The first to nth components in the first clause; n represents the first to the mth components in the second clause; n and m are both positive integers.

[0042] Furthermore, when performing an applicability review on each pieced-together structured natural language statement in the architectural model, it is first necessary to determine whether the pieced-together structured natural language statement contains preset computational rules. For the first structured natural language statement containing preset computational rules, the preset SPARQL SELECT statement is used to review the first structured natural language statement to obtain a first review result. For the second structured natural language statement not containing preset computational rules, the preset SPARQL ASK statement is used to review the second structured natural language statement to obtain a second review result. Finally, the applicability review result of each structured natural language statement is obtained based on the first review result and the second review result.

[0043] Therefore, this application proposes an optimized review method for structured natural language, which solves the problems in related technologies that the review methods are prone to bias in compliance judgment results, cannot accurately locate hidden compliance risks in the design, and are not conducive to staff quickly completing problem rectification and standard verification. It can significantly improve the accuracy of architectural design compliance review.

[0044] Furthermore, to enable those skilled in the art to better understand how the structured natural language application review method of this application ensures the readability and understandability of the application review results, the following description is provided in conjunction with specific embodiments.

[0045] As one possible implementation method, in some embodiments, after performing an applicability review operation on each concatenated structured natural language statement based on the mapping relationship and a preset compliance determination formula to obtain the applicability review result of each structured natural language statement, the method further includes: based on the applicability review result of each structured natural language statement, filtering out target structured natural language statements that are not applicable to review; and storing the review result corresponding to the target structured natural language statement in the corresponding field of the original review result report file.

[0046] Specifically, this application embodiment requires obtaining the applicable review result for each structured natural language statement. The sentence replacement SNL review result report file is read, and based on the mapping relationship between each structured natural language statement and its corresponding concatenated structured natural language statement, the applicable review result for each structured natural language statement (i.e., the original SNL statement) is obtained. Based on the mapping relationship, all applicable review results are stored in memory.

[0047] Furthermore, this embodiment of the application requires writing the applicable review results of each structured natural language statement into a new review result report file. The original review result report file is read to obtain a review result report string, and the applicable review results of each structured natural language statement are sequentially written into the specified positions of the review result report string. Then, a new review result report string containing the applicable review results is written into a specified file, ultimately resulting in a new review result report file containing the applicable review results of each structured natural language statement.

[0048] Therefore, the embodiments of this application can synchronously integrate the determination information of inapplicable SNLs into the original review report, fully retain the review data after fine differentiation, and facilitate the separate extraction of statements that are approved due to inapplicability during subsequent statistics, thereby avoiding the overall review pass rate from being artificially inflated. At the same time, there is no need to reconstruct the original report output logic, and the modification to the existing review engine is small and the compatibility is strong.

[0049] Furthermore, to enable those skilled in the art to better understand the applicable examination method of structured natural language in this application, the following is combined with, for example... Figure 2 and Figure 3 The specific embodiments shown will be described in detail.

[0050] Figure 2 A flowchart illustrating an applicable review method for structured natural language processing provided in this application embodiment.

[0051] like Figure 2 As shown, the applicable review method for structured natural language processing includes the following steps: S201, Review Results Report.

[0052] S202, Output of the file reading and parsing module: Structured Natural Language statements (i.e., SNL statements).

[0053] S203, Clause Replacement Module Output: Concatenated Structured Natural Language Statement (i.e., SNL statement after clause replacement).

[0054] S204, Output of the review engine module: Review results of the concatenated structured natural language statements.

[0055] S205, Output of the results generation module: Applicability review results of structured natural language statements.

[0056] S206, Output module outputs: New review results report file.

[0057] Specifically, Figure 3 This is a schematic diagram illustrating a clause replacement method provided in an embodiment of this application. For example... Figure 3 As shown in the embodiments of this application, an applicability review method for structured natural language statements based on the clause replacement approach is proposed. The essence of applicability review of structured natural language statements is to review the "if" part of the structured natural language statement separately. Therefore, a stable, easy-to-implement and widely applicable approach is to retain the "if" part of the structured natural language statement and replace the "then" part of this structured natural language statement with a clause that is definitely wrong.

[0058] Furthermore, to conduct an applicability review of the structured natural language statement, it is only necessary to call the original review engine to review the concatenated structured natural language statement: if the applicability review result of the concatenated structured natural language statement is non-compliant, it means that there is a target component tuple in the building under review that can meet the requirements of the "if" part of the structured natural language statement, and the structured natural language statement is applicable; otherwise, it means that there is no target component tuple in the building under review that can meet the requirements of the "if" part of the natural language statement, and the structured natural language statement is not applicable.

[0059] For example, suppose the "if" clause in a structured natural language statement A has n clauses, and the constraints of the first clause are as follows: Meanwhile, in the structured natural language statement A, the number of clauses in the "then" part is m, and the constraints of the second clause are as follows: Therefore, the essence of using the existing review engine to perform structured natural language statement application review on the building under review is to search for tuples composed of components that satisfy a preset compliance determination formula within the building under review. If no tuple satisfying the preset compliance determination formula is found, the structured natural language statement A is deemed to pass; otherwise, if at least one tuple satisfying the preset compliance determination formula is found, the structured natural language statement A is deemed to fail. The final non-compliant components are also selected from these tuples.

[0060] Furthermore, the essence of the review of the applicability of structured natural language statements is to determine whether there exists a component plural that satisfies the preset first compliance determination formula in the building under review. If such a plural exists, it means that the "if" part of structured natural language statement A can filter out the target component plural that meets the preset first compliance determination formula, and in this case, structured natural language statement A is applicable; otherwise, structured natural language statement A is not applicable. The preset first compliance determination formula is: ; in, This refers to the first to nth components in the first clause; n is a positive integer.

[0061] Furthermore, after replacing clauses in the structured natural language statement, the number of clauses in the "then" part of the concatenated structured natural language statement is 1. Since the clauses in the "then" part are necessarily incorrect, the essence of using the existing review engine to review the applicability of the concatenated structured natural language statement is to find component tuples in the reviewed building that satisfy the preset second compliance determination formula. Therefore, if the existing review engine determines that the applicability review result of the concatenated structured natural language statement is passed, it means that no component tuple satisfying the preset second compliance determination formula was found in the reviewed building. Since the preset first compliance determination formula and the preset second compliance determination formula are completely equivalent, it can be known that structured natural language statement A is not applicable at this time. If the applicability review result of the concatenated structured natural language statement is failed, it means that there is a component tuple satisfying the preset second compliance determination formula in the reviewed building, and in this case, structured natural language statement A is applicable. The preset second compliance determination formula is: ; in, This refers to the first to nth components in the first clause; n is a positive integer.

[0062] Furthermore, the principle for selecting a suitable, necessarily erroneous clause is as follows: In this embodiment, "then the component has an attribute that does not exist" is ultimately selected as the necessarily erroneous clause, and "component" is replaced with the subject that appears for the first time in the structured natural language statement. For example, as Figure 3 As shown, the first subject in a structured natural language statement is "If the building name includes buildings in the residential section," and the corresponding error-proof clause is "Then the building has an attribute that does not exist." The biggest advantage of using the "has attribute" statement in the structured natural language statement in the error-proof clause is its simplicity and stability. In the above method for generating error-proof clauses, the only variable is the first subject in the structured natural language statement; the rest are concatenated using a fixed structure.

[0063] Therefore, the applicable review method based on clause replacement proposed in this application can reuse the capabilities of review engines in related technologies to achieve efficient, accurate, and applicable review of various structured natural language sentence writing styles.

[0064] Furthermore, to enable those skilled in the art to better understand the applicable examination method of structured natural language in this application, the following description is provided in conjunction with specific embodiments.

[0065] Specifically, the structured natural language application review method of this application can be applied to various intelligent review scenarios for building compliance in the construction field, such as construction drawing review, fire protection review, and pre-opening fire safety inspection. This application's embodiment uses a fire protection review clause actually applied in a certain region as an example for illustration.

[0066] Furthermore, in practical applications, a single specification for intelligent review involves multiple clauses, each clause has multiple review points, and each review point corresponds to multiple structured natural language statements. Taking the execution process corresponding to one of these structured natural language statements as an example, such as... Figure 4 As shown, Figure 4 This application provides a schematic diagram illustrating the application of review results to a single structured natural language statement, as shown in the embodiments of this application. Figure 4The document contains a structured natural language statement (SNR) with the corresponding review point being that fire elevators should have an anteroom. The specific execution steps are as follows: This application first reads and parses the original review result of this project, determines that the subject of the first clause in the current SNR is "building," and replaces the second clause (the "then" conclusion) with the third clause ("then the building has attributes that do not exist") to obtain the corresponding concatenated SNR. The concatenated SNR (and all other concatenated SNRs corresponding to the review results of other clauses) is submitted to the review engine module for review. The concatenated SNR performs an applicable review based on the preset SPARQL SELECT statement. When the review result indicates that the applicable review conditions are not met, the applicable review result is mapped to the original review result based on the mapping relationship, and the applicable review result of each SNR is sequentially written to the specified position in the review result report string.

[0067] Furthermore, based on the examination results of the applicable review in this application embodiment, it was found that although the original examination result of the current structured natural language statement was "passed", the following two situations could not be distinguished based on the original examination result: (1) the current examination item is within the scope of application of the current clause and the design is reasonable, resulting in the examination passing; (2) the current examination item is not within the scope of application of the current clause, and there is no need to perform the examination of this clause, so the examination passes. One beneficial effect of the applicable review method of the structured natural language in this application is that it can distinguish whether a single structured natural language statement is applicable to the examination of this item, avoid blindly expanding the scope of examination, and make the examination conclusion more accurate. For example, as shown in this structured natural language statement and case item: the current item does not have a fire elevator room, and the applicable review result is "condition not passed", indicating that the current item does not need to examine whether the fire elevator room has a vestibule.

[0068] Furthermore, the above steps are an example of the adaptation review process and results for a single structured natural language statement. In a complete project review process, multiple structured natural language statements corresponding to multiple clauses of multiple specifications will be reviewed simultaneously. Another beneficial effect of the structured natural language applicability review method in this application is that it can identify the subset of structured natural language statements that the current project needs to focus on, and eliminate the interference of inapplicable subsets of structured natural language statements on the current intelligent review process. For example, as... Figure 5 As shown, Figure 5This is a schematic diagram of an intelligent adaptive review report provided in this application embodiment. The review of the current project covers 84 structured natural language statements in 11 specifications, but only 22 of them are applicable to the current project situation. The remaining 62 structured natural language statements do not need to be considered. This application embodiment can accurately identify the clauses and structured natural language statements that are applicable to the current project situation, making the review report more accurate and easier to view and review.

[0069] Therefore, the embodiments of this application can filter out effective statements applicable to the structured natural language review method of this application, eliminate inapplicable statements to eliminate invalid interference, reduce the workload of manual review, and improve the accuracy and efficiency of intelligent review as a whole.

[0070] Furthermore, to enable those skilled in the art to better understand the applicable examination method of structured natural language in this application, the following is combined with, for example... Figures 6 to 8 The specific embodiments shown will be described in detail.

[0071] Figure 6 A flowchart illustrating an applicable review optimization method for structured natural language provided in this application embodiment.

[0072] like Figure 6 As shown, the applicable review optimization method for structured natural language processing includes the following steps: S601, Begin.

[0073] S602, Find the subject of the first clause in a structured natural language statement.

[0074] S603 performs clause replacement on structured natural language statements.

[0075] S604, invoke the existing review engine.

[0076] S605. Does the concatenated structured natural language statement contain preset computational rules? If yes, proceed to step S606; otherwise, proceed to step S607.

[0077] S606, invokes the SELECT review logic.

[0078] S607 invokes the ASK-optimized review logic.

[0079] S608, Has the review passed? If yes, proceed to step S609; otherwise, proceed to step S610.

[0080] S609, the output result is not applicable.

[0081] S610, the output results are applicable.

[0082] S611, End.

[0083] Specifically, embodiments of this application propose the following: Figure 7 The proposed time efficiency optimization method for applicability review based on the SPARQL ASK statement aims to adjust the logic of the original review engine, improving its execution time efficiency when performing applicability review tasks. The purpose of inapplicability review is to determine whether a specific component tuple satisfies the "if" clause of a structured natural language statement: if the tuple does not exist, the structured natural language statement is inapplicable; if the tuple exists, the structured natural language statement is applicable. This review logic differs significantly from conventional review. Inapplicability review only focuses on the existence of tuples that meet the requirements, without considering the number of tuples or the specific component information contained within them. For example, a conventional review of the structured natural language statement "If a building has rooms, and the rooms have windows, then the rooms have doors" requires locating specific non-compliant room components. Therefore, in SPARQL, the SELECT syntax is needed to query and select results for specific room components. However, for applicability review, it only needs to determine whether there are (building, room, window) triples that meet the "if" constraint, without considering the number of triples or the specific component information within them.

[0084] Furthermore, when the concatenated structured natural language statement does not contain preset computational rules, the applicability review only needs to determine whether the tuple exists, without querying tuple details, which is highly consistent with the ASK statement in SPARQL. The ASK statement only returns the target pattern as "exists" or "does not exist" in Boolean form, without returning the data of the specific tuple. Therefore, when using the ASK statement, once the SPARQL execution engine finds the first component tuple that satisfies the target pattern, it will immediately terminate the query execution and return a Boolean value of "true", indicating that the target pattern exists; if no target tuple is found after traversing all the data, it returns "false", indicating that the target pattern does not exist. The execution logic of the ASK statement and the SELECT statement are significantly different: SELECT needs to traverse the entire dataset to find all tuples that meet the conditions and return a complete list of target components, which has relatively low execution efficiency; while the "short-circuit execution" characteristic of the ASK statement perfectly matches the core requirement of the applicability review that only needs to determine "exists". The embodiments of this application can use ASK queries to replace the SELECT queries in the original review engine. If the ASK query returns true, it means that the applicability review result of the concatenated structured natural language statement is not passed; otherwise, it is considered that the applicability review is passed.

[0085] Therefore, the embodiments of this application ensure that the applicable review function is compatible with all types of SNL statements. Based on the review task type, SNL statements are mainly divided into pure SPARQL query SNLs and calculation rule SNLs: the former only requires executing the SPARQL query statement, while the latter requires secondary filtering based on the SPARQL query results through a calculation rule framework. The review execution logic of the two types of SNLs differs significantly. Meanwhile, the SNL inapplicability review is an extended function added to the original SNL review logic, which generates additional computational overhead, increasing the overall review process time. The embodiments of this application effectively optimize this step, reduce additional time overhead, and ensure the runtime efficiency of the SNL inapplicability review.

[0086] Furthermore, Figure 8 This is a block diagram illustrating a structured natural language review device provided in an embodiment of this application.

[0087] like Figure 8 As shown, the structured natural language processing device 80 includes: a file reading and parsing module 801, a clause replacement module 802, a review engine module 803, a result generation module 804, and a result output module 805.

[0088] Firstly, the file reading and parsing module 801 is used to read the original review result report file stored in JSON format and obtain all structured natural language statements in the original review result report file. Specifically, this module first reads the JSON string from the original review result report file into memory, then calls a JSON parsing library to parse the JSON string to obtain the review result report object. Secondly, it iterates through all review task result objects stored in the review result report object, extracts the structured natural language statements stored in each review task result object, and collects and stores these structured natural language statements into a structured natural language statement set in memory.

[0089] Secondly, the clause replacement module 802 performs clause replacement operations on each structured natural language statement in the set of structured natural language statements and writes all concatenated structured natural language statements into the clause replacement SNL file. Specifically, this module first traverses the set of structured natural language statements in memory, finds the first subject in each structured natural language statement, and generates a third clause based on this subject. Next, it concatenates the first and third clauses of the structured natural language statements to obtain the concatenated structured natural language statements. Finally, it writes all the concatenated structured natural language statements into the clause replacement SNL file, while simultaneously storing the mapping relationship between each structured natural language statement and its corresponding concatenated structured natural language statement in memory.

[0090] Third, the review engine module 803 is used to review the applicability of concatenated structured natural language statements on a specified architectural semantic model, obtaining the applicability review results (including: pass and fail). Specifically, this module reads the clause replacement SNL file output by the clause replacement module 802, reviews the applicability of all concatenated structured natural language statements stored in the file on the specified architectural model, and stores the applicability review result of each structured natural language statement in the clause replacement SNL review result report file. The review engine module 803 completely reuses the SNL review engine from related technologies, only using SPARQL ASK queries instead of SPARQL SELECT queries used by the review engines in related technologies when executing pure SPARQL query types of concatenated structured natural language statements, to optimize time efficiency. Simultaneously, when compiling the review results, this module determines that structured natural language statements with an ASK query result of true are fail, and SNL statements with an ASK query result of false are pass.

[0091] Fourth, the result generation module 804 is used to read the clause replacement SNL review result report file output by the review engine module 803, and output the applicable review result of each structured natural language statement in memory according to the mapping relationship between each structured natural language statement and its corresponding concatenated structured natural language statement. Specifically, this module reads and parses the clause replacement SNL review result report file, and replaces all the structured natural language statements in it. For example, if the applicable review result of the concatenated structured natural language statement B is "passed", then the structured natural language statement corresponding to the concatenated structured natural language statement B is not applicable. After finding the corresponding structured natural language statement according to the mapping relationship, the result of the inapplicable structured natural language statement is stored in memory. If the review result of the concatenated structured natural language statement C is "failed", then the structured natural language statement corresponding to the concatenated structured natural language statement C is applicable. After finding the corresponding structured natural language statement according to the mapping relationship, the result of the applicable structured natural language statement is stored in memory.

[0092] Fifth, the result output module 805 reads the original review result report file to obtain the review result report string, and sequentially writes the applicable review results of each structured natural language statement output by the result generation module 804 into the specified positions of the review result report string. Finally, the new review result report string containing the applicable review results is written to the specified file to obtain a new review result report file.

[0093] In summary, the Structured Natural Language (SNL) applicability review device proposed in this application can perform SNL statement applicability review without affecting the original review function logic of the SNL review system. The device first reads the original review result report file output by the original SNL review engine, extracts all SNL statements from the original review result report file, then performs applicability review on each SNL statement, and finally writes the applicability review result of each SNL statement into the corresponding position in the review report to determine whether the reviewed building meets the preconditions of the SNL statement. This solves the problem in related technologies where the review results output by SNL review tools are difficult to match the expectations and logic of human drawing review experts, and renders the statistical data of review results meaningless.

[0094] According to the applicable review method for structured natural language proposed in this application, the second clause of each structured natural language statement in the original review result report file can be replaced with a third clause that meets the preset judgment conversion conditions. The first and third clauses of each structured natural language statement are then concatenated to obtain the corresponding concatenated structured natural language statement. Based on a preset compliance judgment formula, an applicable review operation is performed on each concatenated structured natural language statement to obtain the applicable review result for each statement. This solves the problem that review methods in related technologies are prone to bias in compliance judgment results, cannot accurately locate hidden compliance risks in design, and are not conducive to staff quickly completing problem rectification and standard verification. It can significantly improve the accuracy of architectural design compliance review.

[0095] Next, the applicable review apparatus for structured natural language proposed in the embodiments of this application is described with reference to the accompanying drawings.

[0096] Figure 9 This is a block diagram of an applicable review device for structured natural language according to an embodiment of this application.

[0097] like Figure 9 As shown, the applicable review device 10 for structured natural language includes: an acquisition module 100, a replacement module 200, and a review module 300.

[0098] The module includes: an acquisition module 100, which, in response to compliance review requirements, acquires the original review result report file corresponding to the compliance review requirements and obtains a set of structured natural language statements based on the original review result report file; a replacement module 200, which determines the first and second clauses of each structured natural language statement in the set of structured natural language statements, replaces the second clause of each structured natural language statement with a third clause that meets the preset judgment conversion conditions, and concatenates the first and third clauses of each structured natural language statement to obtain the corresponding concatenated structured natural language statement; and an examination module 300, which determines the mapping relationship between each structured natural language statement and each concatenated structured natural language statement, and, based on the mapping relationship and the preset compliance judgment formula, performs applicable examination operations on each concatenated structured natural language statement to obtain the applicable examination result of each structured natural language statement.

[0099] Optionally, in some embodiments, the acquisition module 100 is specifically used to: read the JSON string in the original review result report file; parse the JSON string to obtain a review result report object, and traverse all review task result objects stored in the review result report object to obtain the structured natural language statements stored in each review task result object; and generate a set of structured natural language statements based on the structured natural language statements stored in each review task result object.

[0100] Optionally, in some embodiments, the replacement module 200 is specifically used to: determine the subject that appears for the first time in the first clause of each structured natural language statement; and based on the subject that appears for the first time in the first clause of each structured natural language statement, replace the second clause of the corresponding structured natural language statement with a third clause that satisfies the preset judgment conversion conditions.

[0101] Optionally, in some embodiments, the review module 300 is specifically used for: based on each concatenated structured natural language statement, filtering out a first structured natural language statement containing preset computational rules and a second structured natural language statement not containing preset computational rules; based on a preset compliance determination formula, reviewing the first structured natural language statement using a preset SPARQL SELECT statement to obtain a first review result, and reviewing the second structured natural language statement using a preset SPARQL ASK statement to obtain a second review result; and obtaining an applicable review result for each structured natural language statement based on the first review result and the second review result.

[0102] Optionally, in some embodiments, the preset compliance determination formula is: ; in, The first to nth components in the first clause; n represents the first to the mth components in the second clause; n and m are both positive integers.

[0103] Optionally, in some embodiments, after performing an applicability review operation on each concatenated structured natural language statement based on the mapping relationship and a preset compliance determination formula to obtain the applicability review result of each structured natural language statement, the review module 300 is further configured to: filter out target structured natural language statements that are not applicable to review based on the applicability review result of each structured natural language statement; and store the review result corresponding to the target structured natural language statement in the corresponding field of the original review result report file.

[0104] It should be noted that the foregoing explanation of the applicable review method embodiment for structured natural language also applies to the applicable review device for structured natural language in this embodiment, and will not be repeated here.

[0105] The applicability review device for structured natural language proposed in this application can replace the second clause of each structured natural language statement in the original review result report file with a third clause that meets preset judgment conversion conditions. It then concatenates the first and third clauses of each structured natural language statement to obtain a corresponding concatenated structured natural language statement. Based on a preset compliance judgment formula, it performs an applicability review operation on each concatenated structured natural language statement to obtain the applicability review result for each statement. This solves the problem that review methods in related technologies are prone to bias in compliance judgment results, cannot accurately locate hidden compliance risks in designs, and hinder staff from quickly completing problem rectification and standard verification. It can significantly improve the accuracy of architectural design compliance review.

[0106] Figure 10 A schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device may include: The memory 1001, the processor 1002, and the computer program stored on the memory 1001 and capable of running on the processor 1002.

[0107] When the processor 1002 executes the program, it implements the applicable review method for structured natural language provided in the above embodiments.

[0108] Furthermore, the electronic device also includes: Communication interface 1003 is used for communication between memory 1001 and processor 1002.

[0109] The memory 1001 is used to store computer programs that can run on the processor 1002.

[0110] The memory 1001 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0111] If the memory 1001, processor 1002, and communication interface 1003 are implemented independently, then the communication interface 1003, memory 1001, and processor 1002 can be interconnected via a bus to complete communication between them. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized into address buses, data buses, control buses, etc. For ease of representation, Figure 10 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0112] Optionally, in a specific implementation, if the memory 1001, processor 1002, and communication interface 1003 are integrated on a single chip, then the memory 1001, processor 1002, and communication interface 1003 can communicate with each other through an internal interface.

[0113] The processor 1002 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0114] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the following... Figure 1 The applicable review method for structured natural language is shown.

[0115] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0116] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0117] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0118] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0119] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0120] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0121] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0122] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A method for applicability review of structured natural language processing, characterized in that, include: In response to compliance review requirements, the original review result report file corresponding to the compliance review requirements is obtained, and a set of structured natural language statements is obtained based on the original review result report file; The first clause and the second clause of each structured natural language statement in the set of structured natural language statements are determined, and the second clause of each structured natural language statement is replaced with a third clause that satisfies the preset judgment conversion condition. The corresponding concatenated structured natural language statement is obtained by splicing the first clause and the third clause of each structured natural language statement. The mapping relationship between each structured natural language statement and each concatenated structured natural language statement is determined. Based on the mapping relationship and the preset compliance judgment formula, an applicable review operation is performed on each concatenated structured natural language statement to obtain the applicable review result of each structured natural language statement.

2. The method according to claim 1, characterized in that, The structured natural language statement set obtained based on the original review result report document includes: Read the JSON string from the original review result report file; The JSON string is parsed to obtain a review result report object, and all review task result objects stored in the review result report object are traversed to obtain the structured natural language statements stored in each review task result object; The structured natural language statement set is generated based on the structured natural language statements stored in each review task result object.

3. The method according to claim 1, characterized in that, The step of replacing the second clause of each structured natural language statement with a third clause that satisfies the preset conversion conditions includes: Identify the first subject appearing in the first clause of each structured natural language statement; Based on the first occurrence of the subject in the first clause of each structured natural language statement, the second clause of the corresponding structured natural language statement is replaced with the third clause that satisfies the preset judgment conversion condition.

4. The method according to claim 1, characterized in that, Based on the mapping relationship and the preset compliance determination formula, the applicable review operation is performed on each concatenated structured natural language statement to obtain the applicable review result of each structured natural language statement, including: Based on each concatenated structured natural language statement, a first structured natural language statement containing preset computational rules and a second structured natural language statement not containing the preset computational rules are selected. Based on the preset compliance determination formula, the first structured natural language statement is reviewed using a preset SPARQL SELECT statement to obtain a first review result, and the second structured natural language statement is reviewed using a preset SPARQL ASK statement to obtain a second review result. The applicable review result for each structured natural language statement is obtained based on the first review result and the second review result.

5. The method according to claim 4, characterized in that, The preset compliance determination formula is as follows: ； in, The first to nth components in the first clause; n represents the first to the mth components in the second clause; n and m are both positive integers.

6. The method according to claim 1, characterized in that, After performing an applicability review operation on each concatenated structured natural language statement based on the mapping relationship and the preset compliance determination formula to obtain the applicability review result of each structured natural language statement, the method further includes: Based on the applicable review results of each structured natural language statement, target structured natural language statements that are not applicable to review are selected. The review results corresponding to the target structured natural language statement are stored in the corresponding fields of the original review result report file.

7. A structured natural language application review device, characterized in that, include: The acquisition module is used to respond to compliance review requirements by acquiring the original review result report file corresponding to the compliance review requirements, and obtaining a set of structured natural language statements based on the original review result report file; The replacement module is used to determine the first clause and the second clause of each structured natural language statement in the set of structured natural language statements, and replace the second clause of each structured natural language statement with a third clause that satisfies the preset judgment conversion conditions, and concatenate the first clause and the third clause of each structured natural language statement to obtain the corresponding concatenated structured natural language statement. The review module is used to determine the mapping relationship between each structured natural language statement and each concatenated structured natural language statement, and based on the mapping relationship and the preset compliance judgment formula, to perform applicable review operations on each concatenated structured natural language statement to obtain the applicable review result of each structured natural language statement.

8. The apparatus according to claim 7, characterized in that, The acquisition module is specifically used for: Read the JSON string from the original review result report file; The JSON string is parsed to obtain a review result report object, and all review task result objects stored in the review result report object are traversed to obtain the structured natural language statements stored in each review task result object; The structured natural language statement set is generated based on the structured natural language statements stored in each review task result object.

9. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the program to implement an applicable method for reviewing structured natural language as described in any one of claims 1-6.

10. A computer-readable storage medium storing a computer program, characterized in that, When executed by a processor, the program implements the applicable review method for structured natural language as described in any one of claims 1-6.

Images