A construction industry labor data intelligent tracing method and device and medium

By using pre-trained large language models and OCR technology to perform semantic parsing and compliance verification on labor data in the construction industry, a structured traceability report is generated. This solves the problem of low traceability efficiency caused by inconsistent formats of multi-source heterogeneous data, and enables rapid and accurate location of labor anomalies and evidence preservation.

CN123068599APending Publication Date: 2026-11-10BEIJING BRANCH BUILDING TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611230968.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-14
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Labor data in the construction industry is scattered across multiple heterogeneous data sources, with inconsistent formats and field definitions. This results in low efficiency for manual traceability, difficulty in anomaly identification, and a lack of objective basis for compliance judgment, making it difficult to quickly locate problematic links and responsible parties.

Method used

A pre-trained large language model is used for semantic parsing, entity extraction, and compliance verification. Multi-source heterogeneous labor data is obtained through API interfaces, paper ledgers are processed by OCR recognition technology, structured labor traceability reports are generated, and traceability results are recorded using a blockchain evidence storage system.

Benefits of technology

It achieves semantic unification and standardization of multi-source heterogeneous labor data, automatically locates the links and responsible parties that cause anomalies, shortens the time for troubleshooting anomalies, solves the workload that used to take hours or even days in minutes, and provides an immutable chain of evidence.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123068599A_ABST
    Figure CN123068599A_ABST
Patent Text Reader

Abstract

The application discloses a kind of construction industry labor data intelligent tracing method, device and medium, comprising: in response to labor data tracing request, the multi-source heterogeneous labor data of construction project is obtained, wherein multi-source heterogeneous labor data includes: labor contract data, real-name system record data, attendance data, salary payment record, construction log and paper account image data;Based on pre-training large language model, the semantic analysis, entity extraction and compliance verification of multi-source heterogeneous labor data are carried out, and the tracing state information of labor data is determined, wherein the tracing state information includes: compliance state and abnormal state;Based on tracing state information, structured labor tracing report is generated.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention discloses a method, device, and medium for intelligent traceability of labor data in the construction industry, belonging to the field of intelligent traceability technology for labor data in the construction industry. Background Technology

[0002] In the current construction industry's labor management field, labor data is scattered across various heterogeneous data sources, including labor subcontracting contracts, real-name registration systems, attendance ledgers, wage payment records, construction logs, and paper-based signature ledgers. These data sources have inconsistent formats, inconsistent field definitions, and lack effective cross-source correlation mechanisms. This results in labor data verification and traceability relying entirely on manual verification of multiple documents and experience-based judgment of data rationality. This approach is not only inefficient and labor-intensive, but also suffers from inconsistent review standards among different managers, leading to a lack of objective basis for compliance judgments. Furthermore, the unstructured archiving of paper-based ledger images makes automated retrieval and analysis impossible. In the event of labor disputes or wage arrears, traceability personnel must repeatedly search and compare massive amounts of paper files and spreadsheets manually, making it difficult to quickly locate the problem and accurately identify the responsible party. This severely restricts the timeliness and accuracy of data traceability throughout the entire construction labor employment process.

[0003] There is currently no effective solution to the above problems. Summary of the Invention

[0004] This invention provides a method, device, and medium for intelligent traceability of labor data in the construction industry, which at least solves the technical problems of low efficiency in manual traceability, difficulty in anomaly location, and lack of unified standards for compliance judgment caused by the heterogeneous formats, inconsistent field definitions, diverse semantic expressions, and difficulty in automatically parsing paper ledgers of existing multi-source labor data in the construction industry.

[0005] According to one aspect of the present invention, in order to achieve the above objective, a method for intelligent traceability of labor data in the construction industry is provided, comprising:

[0006] In response to the request for labor data traceability, multi-source heterogeneous labor data of construction projects are obtained. The multi-source heterogeneous labor data includes: labor contract data, real-name registration data, attendance data, wage payment records, construction logs and paper ledger image data.

[0007] Based on a pre-trained large language model, semantic parsing, entity extraction, and compliance verification are performed on multi-source heterogeneous labor data to determine the traceability status information of the labor data. The traceability status information includes compliance status and abnormal status.

[0008] Based on the traceability status information, a structured labor service traceability report is generated.

[0009] Furthermore, acquire multi-source heterogeneous labor data for construction projects, including:

[0010] Electronic labor data is obtained from project management systems, real-name labor registration platforms, and bank payroll systems via API interfaces;

[0011] Obtain batch ledger data in Excel and CSV formats via file upload interface;

[0012] OCR recognition technology is used to extract and structure the text from paper ledger image data and scanned copies of handwritten forms.

[0013] Furthermore, based on a pre-trained large language model, semantic parsing and entity extraction are performed on multi-source heterogeneous labor data, including:

[0014] A pre-trained large language model is used to perform context-free parsing of long texts such as labor contracts and payroll ledgers. The pre-trained large language model is based on the Transformer architecture and has a context window of no less than 100K tokens.

[0015] A multimodal visual language model was used to perform visual semantic analysis and text reconstruction on paper ledger image data;

[0016] Based on the structured output interface of the pre-trained large language model, a custom labor traceability field template is provided, which automatically extracts personnel entities, work group entities, project entities, working hours entities, salary entities, and work sequence entities.

[0017] The extracted entities are semantically normalized.

[0018] Furthermore, based on a pre-trained large language model, compliance verification is performed on multi-source heterogeneous labor data, including:

[0019] The text of the labor management regulations for the housing and construction industry is converted into a high-dimensional vector using a text encoder, and a vector database is constructed.

[0020] During the inference process of the pre-trained large language model, the extracted entity data is encoded into vectors, and the most relevant normative clauses are retrieved from the vector database using the approximate nearest neighbor algorithm.

[0021] The retrieved regulatory clauses are combined with entity data as additional context to form prompt words, which are then input into a pre-trained large language model for compliance judgment, identifying issues such as missing employment registration, abnormal working hours, irregular salary payment, and non-compliant personnel operations.

[0022] Furthermore, based on the traceability status information, a structured labor service traceability report is generated, including:

[0023] When the traceability status information is compliant, a compliance confirmation report is generated;

[0024] When the traceability status information is abnormal, determine the abnormality type, the abnormality occurrence process, the responsible party, and the time period of the violation, and generate a traceability report that includes the abnormality source tracing conclusions and rectification suggestions.

[0025] Furthermore, the type of anomaly, the stage in which it occurred, the responsible party, and the time period of the violation should be determined, including:

[0026] By leveraging the logical reasoning capabilities of pre-trained large language models, we can connect full-time-series data on personnel entry, daily operations, attendance statistics, salary settlement, wage payment, and personnel departure to construct a multi-dimensional traceability graph of personnel, work teams, projects, working hours, wages, and compliance records.

[0027] Based on the multi-dimensional source map, cross-comparison of multi-source data and temporal logic deduction are performed to locate abnormal data nodes and responsible parties.

[0028] Furthermore, after generating a structured labor service traceability report based on the traceability status information, it also includes:

[0029] The structured labor service traceability report, original data correlation, and inference logs of the pre-trained large language model are synchronized to the blockchain evidence storage system to generate tamper-proof evidence records. These records are used for evidence in labor disputes, regulatory verification, and project review.

[0030] Furthermore, in response to a labor data traceability request, after obtaining multi-source heterogeneous labor data for the construction project, the process also includes:

[0031] Based on the time-series characteristics of historical labor data and corresponding violation event tags in historical labor traceability records, a violation risk prediction model is constructed using a time-series prediction model. The violation risk prediction model is used to predict the probability and type of labor violation risk in the current project within a preset future time period.

[0032] When the probability of violation exceeds the preset risk threshold, an early warning message is generated and pushed to the regulatory terminal.

[0033] According to one embodiment of the present invention, a smart traceability device for labor data in the construction industry is also provided, comprising:

[0034] The acquisition module is used to respond to labor data traceability requests and acquire multi-source heterogeneous labor data of construction projects. The multi-source heterogeneous labor data includes: labor contract data, real-name registration data, attendance data, wage payment records, construction logs and paper ledger image data.

[0035] The processing module is used to perform semantic parsing, entity extraction and compliance verification on multi-source heterogeneous labor data based on a pre-trained large language model, and to determine the traceability status information of the labor data, including compliance status and abnormal status.

[0036] The generation module is used to generate structured labor traceability reports based on traceability status information.

[0037] According to another aspect of the present invention, a computer-readable storage medium is also provided, on which a computer program is stored, which, when executed by a processor, implements the intelligent traceability method for construction industry labor data according to various embodiments of the present invention.

[0038] In this embodiment of the invention, a pre-trained large language model is used to perform full-volume lossless parsing of long texts such as labor contracts and wage ledgers, and multimodal semantic recognition of paper ledger images. Combined with the RAG retrieval enhancement mechanism and the linkage verification with the housing and construction labor standard knowledge base, the semantic unification and standardization of multi-source heterogeneous labor data are achieved. This avoids the problem of missed checks and judgments caused by differences in terminology and inconsistent field formats, and significantly reduces the workload of manually checking multiple documents one by one and subjective judgment bias. On this basis, a multi-dimensional traceability map is constructed through cross-source logical reasoning and temporal correlation analysis of the large model. It automatically locates the abnormal generation link, responsible party and violation period, and generates a structured traceability report containing rectification suggestions. This shortens the investigation time of labor abnormality issues from the traditional hours or even days to minutes. At the same time, it realizes the standardized output of traceability results and blockchain evidence storage, providing an immutable and complete chain of evidence for labor dispute evidence and regulatory verification. Attached Figure Description

[0039] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0040] Figure 1 This is a flowchart of a method for intelligent traceability of labor data in the construction industry according to one embodiment of the present invention;

[0041] Figure 2 This is a structural block diagram of a smart traceability device for labor data in the construction industry according to one embodiment of the present invention. Detailed Implementation

[0042] To enable those skilled in the art to better understand the present invention, 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. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0043] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, apparatus, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0044] According to an embodiment of the present invention, an embodiment of an intelligent traceability method for labor data in the construction industry is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer device such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0045] This method embodiment can be executed in an electronic device or similar computing device that includes a memory and a processor. Taking data processing as an example, the data processing may include one or more processors (processors may include, but are not limited to, central processing units (CPUs), graphics processing units (GPUs), digital signal processing (DSP) chips, microcontroller units (MCUs), field-programmable gate arrays (FPGAs), neural network processors (NPUs), tensor processors (TPUs), artificial intelligence (AI) type processors, etc.) and a memory for storing data. Optionally, the above data processing may also include transmission devices, input / output devices, and display devices for communication functions. Those skilled in the art will understand that the above structural description is merely illustrative and does not limit the structure of the above data processing. For example, the data processing may include more or fewer components than described above, or have a different configuration than described above.

[0046] The memory can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the intelligent traceability method for construction industry labor data in this embodiment of the invention. The processor executes various functional applications and data processing by running the computer program stored in the memory, thereby realizing the aforementioned intelligent traceability method for construction industry labor data. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to the mobile terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0047] The transmission device is used to receive or send data via a network. Specific examples of the network mentioned above may include a wireless network provided by the mobile terminal's communication provider. In one example, the transmission device includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device may be a Radio Frequency (RF) module, used for wireless communication with the Internet.

[0048] Display devices can be, for example, touchscreen liquid crystal displays (LCDs) and touch displays (also referred to as "touchscreens" or "touch displays"). The LCD allows users to interact with the user interface of the mobile terminal. In some embodiments, the mobile terminal has a graphical user interface (GUI), which allows users to interact with the GUI through finger contact and / or gestures on a touch-sensitive surface. Optional human-computer interaction functions include: creating web pages, drawing, word processing, creating electronic documents, playing games, video conferencing, instant messaging, sending and receiving emails, call interfaces, playing digital video, playing digital music, and / or web browsing, etc. Executable instructions for performing the above human-computer interaction functions are configured / stored in one or more processor-executable computer program products or readable storage media.

[0049] Figure 1 According to one embodiment of the present invention, a method for intelligent traceability of labor data in the construction industry includes:

[0050] Step S110: In response to the labor data traceability request, obtain multi-source heterogeneous labor data for the construction project. This multi-source heterogeneous labor data includes: labor contract data, real-name registration data, attendance data, wage payment records, construction logs, and image data of paper ledgers. Specific details are as follows:

[0051] In step S110, when a regulatory authority, project manager, or labor dispute mediator triggers a labor tracing operation, the system receives a labor data tracing request. This request can be an instruction generated by clicking the "Trace" button on the system interface, or a tracing instruction passed through an API interface carrying filter conditions such as project number, work group name, personnel ID number, or time period. Upon receiving the request, the system immediately initiates a multi-source heterogeneous labor data collection process for the target construction project to obtain all labor data related to this tracing request.

[0052] Specifically, the multi-source heterogeneous labor data includes at least labor contract data, real-name registration data, attendance data, wage payment records, construction logs, and paper ledger image data. The data includes labor contract information such as the contracting parties, contract period, job type, and salary terms in labor subcontracting contracts, labor contracts, and labor agreements; real-name registration data includes worker identity information, entry registration time, work team, job qualification certificate information, and exit records collected from the construction worker real-name management platform; attendance data covers daily attendance records, attendance summary tables, and access control records, including fields such as attendance days, absence days, overtime hours, and leave records; wage payment records include monthly wage ledgers, bank transfer statements, and wage signature confirmation forms, covering basic salary, performance salary, overtime pay, actual amount paid, and payment time; construction logs record daily construction work content, worker arrangements, and work team divisions; and paper ledger image data includes unstructured materials in the form of images or scanned copies, such as handwritten paper attendance sheets, paper wage signature ledgers, entry safety education signature forms, and public notice photos.

[0053] The acquisition of the aforementioned multi-source heterogeneous labor data is achieved in parallel through multiple acquisition channels. For electronic data stored in various business systems, the system interfaces with project management systems, labor real-name registration platforms, bank payroll systems, and attendance terminal systems via standard APIs. It automatically retrieves the target project's master labor contract data, real-name registration information, daily attendance records, and payroll transaction logs according to preset acquisition cycles or time range parameters in traceability requests. For batch ledger files in Excel and CSV formats, the system provides a file upload interface, supporting manual uploads or scheduled synchronization to import data. It also automatically parses and structures the tabular data using preset field mapping templates. For paper ledger image data, the system supports uploading various image formats, including on-site photos and scanned PDFs. It uses an OCR recognition engine to locate, segment, and extract text from the images, converting unstructured image data into computer-processable text information. Finally, all data is organized according to a unified data model, providing complete data input for subsequent semantic analysis of large-scale models.

[0054] Step S120: Based on the pre-trained large language model, perform semantic parsing, entity extraction, and compliance verification on the multi-source heterogeneous labor data to determine the traceability status information of the labor data. The traceability status information includes compliance status and abnormal status, as detailed below:

[0055] In step S120, after completing the collection and preprocessing of multi-source heterogeneous labor data for construction projects, the system inputs all standardized and organized data into a pre-trained large language model for deep semantic analysis and intelligent verification. Since construction industry labor data originates from different business systems, there are significant differences in the naming methods, field definitions, and expression habits of each system for the same business entity. For example, "attendance days" in the attendance system, "actual attendance" in the payroll, and "actual number of workers" in the construction log may refer to the same concept but use different expressions. If only traditional keyword matching or rule engines are used for processing, it is easy to cause deviations in semantic understanding and omissions in data association. This invention introduces a pre-trained large language model based on the Transformer architecture as the core semantic understanding engine. This model has been fully pre-trained on massive amounts of Chinese government documents, engineering construction specifications, and industry corpora, possessing native semantic understanding capabilities for professional terms, standard texts, and industry-standard expressions in the construction labor field. It can achieve unified understanding and alignment of multi-source heterogeneous labor data at the semantic level without having to write complex parsing rules for each type of data source.

[0056] Specifically, in the semantic parsing stage, the pre-trained large language model leverages its ultra-long context window capacity to perform lossless loading and overall semantic modeling of extremely long texts, often tens of thousands of words long, such as labor contracts, payroll ledgers, and construction logs. This avoids the problems of key context loss and broken semantic connections across paragraphs caused by text truncation in traditional methods. When processing labor contracts, the model can comprehensively understand the contracting parties, work period agreements, and salary calculation methods by combining the context of the contract, rather than identifying individual fields in isolation. When processing paper ledger image data, the system simultaneously calls a multimodal visual language model to jointly recognize printed and handwritten text in the images, and combines layout structure analysis to restore the row and column correspondences in the tables, converting unstructured image information into structured text with semantic annotations. Based on its semantic understanding capabilities of various text styles, including Chinese tables, paragraphs, and clauses, the model extracts semantically complete and context-related sets of labor information from multiple documents of different sources and formats, effectively eliminating semantic gaps caused by data fragmentation and providing high-quality semantic feature input for subsequent entity extraction.

[0057] During the entity extraction phase, the system, based on the structured output interface of a pre-trained large language model, pre-defines standardized field templates required for labor traceability. These templates cover core traceability dimensions such as personnel entities (worker name, ID number, job type, skill level), work group entities (work group name, work group leader, subcontractor), project entities (project number, contract name, construction location), working hours entities (statistical period, attendance days, overtime hours, leave days), salary entities (basic salary, performance-based salary, actual amount paid, payment date, bank statement number), and work sequence entities (entry date, exit date, work content, safety briefing records). The large language model matches and maps the semantic feature vectors obtained in the semantic parsing phase with the aforementioned custom field templates, automatically extracting each entity and its attributes in key-value pairs, and aggregating and reorganizing fragmented information scattered across different documents according to a unified data dimension. In scenarios where the same entity has multiple aliases or non-standard entries in different documents, the model automatically performs semantic normalization processing by using its ability to judge the semantic similarity of Chinese. For example, it unifies different expressions such as "Wang Gong", "Wang XX", and "Project Manager Wang Mou" to the same personnel entity, and unifies fields such as "attendance due", "attendance days", and "number of clock-ins" into the attendance statistics dimension under the working hours entity. This forms a standardized entity network with personnel, work groups, projects, working hours, salaries, and time series as core nodes, and establishes a data foundation with unified field definitions and standardized data types for subsequent compliance verification and source tracing reasoning.

[0058] During the compliance verification phase, the system integrates a retrieval-enhanced generation (RAG) architecture based on a pre-trained large language model. It converts national and local housing and construction labor management regulations, such as the "Measures for the Real-Name Management of Construction Workers," the "Regulations on Guaranteeing the Payment of Wages to Migrant Workers," and the "Interim Measures for the Management of Special Accounts for Migrant Workers' Wages in the Engineering Construction Field," into high-dimensional vectors using a text encoder, constructing a domain-specific regulation vector database. After the model completes entity extraction from the current project's labor data, the system encodes the extracted key entity information and project attributes into a query vector. Using an approximate nearest neighbor algorithm, it retrieves the most relevant and highly matched regulatory clauses from the regulation vector database. These retrieved clauses, as additional context, are combined with the extracted entity data and the original traceability request to form prompt words, which are then input into the pre-trained large language model for in-depth compliance reasoning and judgment. The model integrates the matching relationship between attendance hours and wage payment amount, the logical consistency between personnel entry time and construction log recording time period, the correspondence between the number of registered team members and the actual number of attendees, the degree of conformity between wage payment time and contractual period, and whether special operation personnel have the corresponding qualifications. It automatically identifies irregular and abnormal scenarios in the labor employment process, such as missing employment registration, discrepancies between attendance and salary, delayed wage payment, personnel working without certificates, and failure to register personnel information in a timely manner.

[0059] Finally, the pre-trained large language model determines the traceability status information of the labor data based on the comprehensive results of compliance verification. This traceability status information includes two judgment conclusions: compliance status and abnormal status. When all verification dimensions pass and no abnormalities are found, the model determines the traceability status information as compliant; when any verification dimension triggers violation identification or there is a data conflict that cannot pass logical consistency, the model determines the traceability status information as abnormal. For abnormal statuses, the model simultaneously outputs the specific abnormality type, the violated regulatory clauses, and the project stage where the abnormality occurred, providing accurate problem location basis for subsequent generation of abnormality tracing conclusions and rectification suggestions.

[0060] Step S140: Based on the traceability status information, generate a structured labor traceability report, the specific content of which is as follows:

[0061] In step S140, after the pre-trained large language model completes semantic parsing, entity extraction, and compliance verification of multi-source heterogeneous labor data, and determines the traceability status information of the labor data, the system automatically triggers the generation process of a structured labor traceability report based on this traceability status information. Traditional labor data traceability output is usually in the form of manually compiled text descriptions or scattered table screenshots, which are not only inconsistent in format and incomplete in content, but also difficult to be directly parsed and utilized by other business systems, resulting in the inability to achieve standardized retention and cross-system sharing of traceability results. This invention, through the structured output interface of the pre-trained large language model, automatically assembles the traceability status information and all associated verification evidence, data basis, and reasoning process according to a predefined standardized template, generating a structured labor traceability report that can be used for both manual review and machine parsing. This fundamentally solves the problems of arbitrary format, fragmented content, and non-programmable utilization of traditional traceability reports.

[0062] Specifically, the structured labor service traceability report includes at least five core components: basic project information, a traceability results overview, compliance verification details, an anomaly details list, and rectification suggestions. The basic project information section records the project number, project name, bid section name, traceability time period, triggering reason, and traceability request initiator for this traceability, providing a complete background identifier for the report. The traceability results overview section presents the overall conclusion of this traceability in a concise manner, indicating whether the traceability status is compliant or abnormal, and includes a confidence score from the model for this conclusion, enabling supervisors or project managers to immediately understand the core conclusions of this traceability.

[0063] When the traceability status information is in compliance status, the system calls the pre-trained large language model to summarize and organize the passing results of all verification dimensions, listing the relevant regulatory clauses and corresponding verification passing data for each verification dimension in the form of a positive list, and generating a compliance confirmation report. The compliance confirmation report also includes the index of all original data sources involved in this traceability and the verification timestamps, ensuring that the compliance conclusion has complete evidentiary support.

[0064] When the traceability status information is in an abnormal state, the system focuses on expanding the abnormal details list in the report. Based on the violation types and risk tags identified during the compliance verification phase, the pre-trained large language model, combined with data anomaly nodes, temporal logic breakpoints, and cross-source data conflict records extracted from the multi-dimensional traceability graph, automatically categorizes and lists all abnormal events discovered in this traceability. Each abnormal event includes an abnormality type tag (such as missing employment registration, discrepancies between attendance and wages, overdue wage payments, unlicensed work, and untimely personnel information registration), information on the specific personnel or work groups involved, the project location or work period where the abnormality occurred, the specific regulatory clause number and a summary of the original clause, and the key evidence chain used by the model to determine the abnormality, i.e., which specific data sources conflicted or were missing, leading to the identification of the abnormality. The system also utilizes the logical reasoning and attribution analysis capabilities of pre-trained large language models to conduct in-depth causal tracing of each abnormal event, identifying the link in the chain of events that caused the abnormality (e.g., whether it was due to the labor subcontractor's failure to report personnel change information in a timely manner, resulting in missing records, or due to the malfunction of attendance equipment causing inaccurate work hour statistics, or due to calculation errors in the payroll process leading to a discrepancy between the actual amount paid and the amount due), accurately locating the work teams, subcontractors, and corresponding management personnel involved in the abnormality, and clarifying the specific time period in which the abnormality occurred. This allows managers to directly obtain a complete profile of the abnormal event without having to manually search through massive amounts of original documents.

[0065] After identifying and attributing abnormal events, the system further utilizes the suggestion generation capabilities of a pre-trained large language model. Combining the type, severity, and violated regulations of the abnormal event, it matches or dynamically generates targeted rectification suggestions from the rectification measure library. These suggestions include specific actions such as immediately ceasing the illegal operation, submitting missing registration information within a specified timeframe, recalculating and issuing back wages, and arranging training and assessment for unlicensed personnel to obtain the corresponding qualification certificates. The system also includes suggested completion deadlines and a list of materials requiring resubmission for review after rectification. This standardized and actionable approach provides managers with clear rectification guidelines, enabling the traceability report to not only identify and locate problems but also guide the closed-loop rectification process.

[0066] Finally, the system organizes all the above content according to a predefined JSON Schema structure, generating a standardized labor service traceability report that includes both human-readable natural language descriptions and machine-parsable structured fields. This report is output in JSON format, adapting to the needs of automatic database archiving in the labor service traceability system, data reporting to the regulatory platform, and calls to third-party business systems. Simultaneously, the system packages the generated report together with the original data relationships involved in this traceability, the retrieval records of regulatory clauses used in the large-scale model inference process, and the inference chain logs, and pushes this package to the blockchain evidence storage system to generate tamper-proof evidence records. This ensures that every conclusion and every abnormal event recorded in the traceability report is verifiable and traceable, and can withstand full-process verification in labor dispute evidence collection, administrative regulatory verification, and internal project review audits.

[0067] Based on steps S110 to S140 above, in this embodiment of the invention, a pre-trained large language model is used to perform full-volume lossless parsing of long texts such as labor contracts and wage ledgers, and multimodal semantic recognition of paper ledger images. Combined with the RAG retrieval enhancement mechanism and the linkage verification with the housing and construction labor standard knowledge base, the semantic unification and standardization of multi-source heterogeneous labor data are achieved. This avoids the problem of missed checks and judgments caused by differences in terminology and inconsistent field formats, and significantly reduces the workload of manually checking multiple documents one by one and subjective judgment bias. On this basis, a multi-dimensional traceability map is constructed through cross-source logical reasoning and temporal correlation analysis of the large model, which automatically locates the abnormal generation link, responsible party and violation period and generates a structured traceability report containing rectification suggestions. This shortens the investigation time of labor abnormality issues from the traditional hours or even days to minutes. At the same time, it realizes the standardized output of traceability results and blockchain evidence storage, providing an immutable and complete chain of evidence for labor dispute evidence and regulatory verification.

[0068] The method of embodiments of the present invention for obtaining multi-source heterogeneous labor data of construction projects includes: obtaining electronic labor data from project management systems, real-name labor registration platforms, and bank payroll systems through API interfaces; obtaining batch ledger data in Excel and CSV formats through file upload interfaces; and extracting and converting text from paper ledger image data and scanned copies of handwritten forms using OCR recognition technology.

[0069] Furthermore, based on a pre-trained large language model, semantic parsing and entity extraction are performed on multi-source heterogeneous labor data. This includes: using a pre-trained large language model to perform context-free parsing of long texts of labor contracts and payroll ledgers; the pre-trained large language model is based on the Transformer architecture and has a context window of no less than 100K tokens; using a multimodal visual language model to perform visual semantic parsing and text reconstruction of paper ledger image data; based on the structured output interface of the pre-trained large language model, custom labor traceability field templates are defined to automatically extract personnel entities, work group entities, project entities, working hours entities, salary entities, and work sequence entities; and semantic normalization is performed on the extracted entities.

[0070] Furthermore, based on a pre-trained large language model, compliance verification is performed on multi-source heterogeneous labor data. This includes: converting the text of labor management regulations for the housing and construction industry into high-dimensional vectors using a text encoder and constructing a vector database; during inference in the pre-trained large language model, encoding the extracted entity data into vectors and retrieving the most relevant regulatory clauses from the vector database using an approximate nearest neighbor algorithm; combining the retrieved regulatory clauses as additional context with the entity data to form prompt words, inputting them into the pre-trained large language model for compliance judgment, and identifying issues such as missing employment registration, abnormal working hours, irregular salary payments, and non-compliant personnel operations.

[0071] Furthermore, based on the traceability status information, a structured labor service traceability report is generated, including: when the traceability status information is in a compliant state, a compliance confirmation report is generated; when the traceability status information is in an abnormal state, the abnormality type, the abnormality generation process, the responsible party, and the violation period are determined, and a traceability report containing the abnormality source tracing conclusion and rectification suggestions is generated.

[0072] Furthermore, the anomaly type, the stage where the anomaly occurred, the responsible party, and the time period of the violation are determined. This includes: utilizing the logical reasoning capabilities of a pre-trained large language model to correlate full-time-series data on personnel entry, daily operations, attendance statistics, payroll settlement, wage payment, and personnel departure, and constructing a multi-dimensional traceability map of personnel-team-project-working hours-wage-compliance records; and based on the multi-dimensional traceability map, performing cross-comparison of multi-source data and temporal logical deduction to locate data anomaly nodes and responsible parties.

[0073] Furthermore, after generating a structured labor service traceability report based on the traceability status information, the process also includes: synchronizing the structured labor service traceability report, the original data association relationship, and the inference log of the pre-trained large language model to the blockchain evidence storage system to generate an immutable evidence storage record. The evidence storage record is used for evidence in labor disputes, regulatory verification, and project review.

[0074] Furthermore, in response to the labor data traceability request, after obtaining multi-source heterogeneous labor data of the construction project, the process also includes: based on the time-series characteristics of historical labor data in the historical labor traceability records and the corresponding violation event tags, constructing a violation risk prediction model using a time-series prediction model. The violation risk prediction model is used to predict the probability and type of labor violation risk of the current project in a preset future period. When the probability of violation risk exceeds the preset risk threshold, an active early warning information is generated and pushed to the regulatory terminal.

[0075] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0076] This invention also provides an intelligent traceability device for labor data in the construction industry, which is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0077] Figure 2 According to one embodiment of the present invention, a smart traceability device for labor data in the construction industry includes:

[0078] The acquisition module 201 is used to respond to the labor data traceability request and acquire multi-source heterogeneous labor data of the construction project. The multi-source heterogeneous labor data includes: labor contract data, real-name registration data, attendance data, wage payment records, construction logs and paper ledger image data.

[0079] Processing module 202 is used to perform semantic parsing, entity extraction and compliance verification on multi-source heterogeneous labor data based on a pre-trained large language model, and to determine the traceability status information of the labor data, wherein the traceability status information includes: compliance status and abnormal status.

[0080] The generation module 203 is used to generate a structured labor traceability report based on the traceability status information.

[0081] It should be noted that the above modules can be implemented by software or hardware. For the latter, they can be implemented in the following ways, but are not limited to: all the above modules are located in the same processor; or, the above modules are located in different processors in any combination.

[0082] According to one embodiment of the present invention, a computer-readable storage medium is also provided, on which a computer program is stored, which, when executed by a processor, implements the above-described intelligent traceability method for labor data in the construction industry.

[0083] Optionally, in this embodiment, the processor can be configured to perform the following steps via a computer program:

[0084] Step S1: In response to the labor data traceability request, obtain multi-source heterogeneous labor data of the construction project. The multi-source heterogeneous labor data includes: labor contract data, real-name registration data, attendance data, wage payment records, construction logs and paper ledger image data.

[0085] Step S2: Based on the pre-trained large language model, perform semantic parsing, entity extraction and compliance verification on multi-source heterogeneous labor data to determine the traceability status information of the labor data. The traceability status information includes compliance status and abnormal status.

[0086] Step S3: Generate a structured labor service traceability report based on the traceability status information.

[0087] According to one embodiment of the present invention, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the storage medium is located to execute the above-described intelligent traceability method for construction industry labor data.

[0088] Optionally, in this embodiment, the storage medium may be configured to store a computer program for performing the following steps:

[0089] Step S1: In response to the labor data traceability request, obtain multi-source heterogeneous labor data of the construction project. The multi-source heterogeneous labor data includes: labor contract data, real-name registration data, attendance data, wage payment records, construction logs and paper ledger image data.

[0090] Step S2: Based on the pre-trained large language model, perform semantic parsing, entity extraction and compliance verification on multi-source heterogeneous labor data to determine the traceability status information of the labor data. The traceability status information includes compliance status and abnormal status.

[0091] Step S3: Generate a structured labor service traceability report based on the traceability status information.

[0092] Optionally, in this embodiment, the storage medium may include, but is not limited to, various media capable of storing computer programs, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0093] According to one embodiment of the present invention, a computer program product is also provided, including a computer program that, when executed by a processor, implements the above-described intelligent traceability method for labor data in the construction industry.

[0094] Optionally, in this embodiment, the above-mentioned computer program product can be configured as a computer program that performs the following steps:

[0095] Step S1: In response to the labor data traceability request, obtain multi-source heterogeneous labor data of the construction project. The multi-source heterogeneous labor data includes: labor contract data, real-name registration data, attendance data, wage payment records, construction logs and paper ledger image data.

[0096] Step S2: Based on the pre-trained large language model, perform semantic parsing, entity extraction and compliance verification on multi-source heterogeneous labor data to determine the traceability status information of the labor data. The traceability status information includes compliance status and abnormal status.

[0097] Step S3: Generate a structured labor service traceability report based on the traceability status information.

[0098] Optionally, specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementations, and will not be repeated here.

[0099] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0100] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between units or modules may be electrical or other forms.

[0101] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0102] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0103] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0104] The above are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for intelligent traceability of labor data in the construction industry, characterized in that, include: In response to a labor data traceability request, multi-source heterogeneous labor data of the construction project is obtained, wherein the multi-source heterogeneous labor data includes: labor contract data, real-name registration data, attendance data, wage payment records, construction logs and paper ledger image data; Based on a pre-trained large language model, semantic parsing, entity extraction, and compliance verification are performed on the multi-source heterogeneous labor data to determine the traceability status information of the labor data, wherein the traceability status information includes: compliance status and abnormal status. Based on the traceability status information, a structured labor traceability report is generated.

2. The intelligent traceability method for construction industry labor data according to claim 1, characterized in that, Acquire multi-source heterogeneous labor data for construction projects, including: Electronic labor data is obtained from project management systems, real-name labor registration platforms, and bank payroll systems via API interfaces; Obtain batch ledger data in Excel and CSV formats via file upload interface; OCR recognition technology is used to extract and structure the text from paper ledger image data and scanned copies of handwritten forms.

3. The intelligent traceability method for labor data in the construction industry according to claim 2, characterized in that, Based on a pre-trained large language model, semantic parsing and entity extraction are performed on the multi-source heterogeneous labor data, including: A pre-trained large language model is used to perform context-free parsing of long texts such as labor contracts and payroll ledgers. The pre-trained large language model is based on the Transformer architecture and has a context window of no less than 100K tokens. A multimodal visual language model was used to perform visual semantic analysis and text reconstruction on paper ledger image data; Based on the structured output interface of the pre-trained large language model, a custom labor traceability field template is created to automatically extract personnel entities, work group entities, project entities, working hours entities, salary entities, and work sequence entities. The extracted entities are semantically normalized.

4. The intelligent traceability method for labor data in the construction industry according to claim 3, characterized in that, Based on a pre-trained large language model, compliance verification is performed on the multi-source heterogeneous labor data, including: The text of the labor management regulations for the housing and construction industry is converted into a high-dimensional vector using a text encoder, and a vector database is constructed. During the inference process of the pre-trained large language model, the extracted entity data is encoded into vectors, and the most relevant normative clauses are retrieved from the vector database using an approximate nearest neighbor algorithm. The retrieved regulatory clauses are combined with the entity data as additional context to form prompt words, which are then input into the pre-trained large language model for compliance determination, identifying issues such as missing employment registration, abnormal working hours, irregular salary payment, and non-compliant personnel operations.

5. The intelligent traceability method for labor data in the construction industry according to claim 1, characterized in that, Based on the traceability status information, a structured labor service traceability report is generated, including: When the traceability status information is in a compliant state, a compliance confirmation report is generated; When the traceability status information is in an abnormal state, the abnormality type, the abnormality generation process, the responsible party, and the violation period are determined, and a traceability report containing the abnormality source tracing conclusion and rectification suggestions is generated.

6. The intelligent traceability method for labor data in the construction industry according to claim 5, characterized in that, Determine the type of anomaly, the stage in which the anomaly occurred, the responsible party, and the time period of the violation, including: By leveraging the logical reasoning capabilities of the pre-trained large language model, the full-time-series data of personnel entry, daily operations, attendance statistics, salary settlement, wage payment, and personnel departure are linked to construct a multi-dimensional traceability graph of personnel-team-project-working hours-wages-compliance records; Based on the multidimensional source map, cross-comparison of multi-source data and temporal logic deduction are performed to locate abnormal data nodes and responsible parties.

7. The intelligent traceability method for labor data in the construction industry according to claim 1, characterized in that, After generating the structured labor service traceability report based on the traceability status information, the process further includes: The structured labor service traceability report, the original data association, and the inference log of the pre-trained large language model are synchronized to the blockchain evidence storage system to generate an immutable evidence record. The evidence record is used for evidence collection in labor disputes, regulatory verification, and project review.

8. The intelligent traceability method for labor data in the construction industry according to claim 1, characterized in that, In response to a labor data traceability request, after obtaining multi-source heterogeneous labor data for the construction project, the following is also included: Based on the time-series characteristics of historical labor data and corresponding violation event tags in historical labor traceability records, a violation risk prediction model is constructed using a time-series prediction model. The violation risk prediction model is used to predict the probability and type of labor violation risk in the current project within a preset future time period. When the probability of violation exceeds a preset risk threshold, an early warning message is generated and pushed to the regulatory terminal.

9. A smart traceability device for labor data in the construction industry, characterized in that, include: The acquisition module is used to respond to labor data traceability requests and acquire multi-source heterogeneous labor data of construction projects. The multi-source heterogeneous labor data includes: labor contract data, real-name registration data, attendance data, wage payment records, construction logs and paper ledger image data. The processing module is used to perform semantic parsing, entity extraction and compliance verification on the multi-source heterogeneous labor data based on a pre-trained large language model, and determine the traceability status information of the labor data, wherein the traceability status information includes: compliance status and abnormal status; The generation module is used to generate a structured labor traceability report based on the traceability status information.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored executable program, wherein, when the executable program is executed, it controls the device on which the computer-readable storage medium is located to perform the intelligent traceability method for construction industry labor data as described in any one of claims 1 to 8.