A Smart Verification System and Method for Test Reports

Abstract

This invention discloses an intelligent verification system and method for inspection reports. The system includes: a verification task management module, used to send task data for initiating tasks via API calls and save tasks to the local system; a prompt word management module, divided into prompt word common part management and inspection item management, used for supplementing verification and expanding associations of anchor keywords, and for point-to-point precise verification of specific inspection items, respectively; an inspection result management module, which obtains tasks from the local system and reconstructs the information value of inspection reports according to usage scenarios; and a verification task interface module, used to adjust the text content for verification into code structure and display it on the front-end page. This invention can build a full-dimensional verification system framework covering general scenarios and subdivided fields, solving common problems in existing technologies such as multi-source data alignment and dynamic rule adaptation.

Description

Technical Field

[0001] This invention relates to the field of test report data management technology, specifically to an intelligent verification system and method for test reports. Background Technology

[0002] In the field of laboratory testing, Laboratory Information Management System (LIMS) serves as a core business platform and has been widely applied to key aspects such as testing task allocation, sample lifecycle management, experimental data recording, and test report generation.

[0003] The LIMS system has achieved standardization and traceability of the testing process through information technology, providing basic support for laboratory quality control and compliance. As the volume of testing business continues to grow and the requirements for data accuracy become higher, the LIMS system is gradually integrating more automated and intelligent functions to improve overall operational efficiency and data reliability.

[0004] Under the existing LIMS system architecture, the verification of test reports and original records usually relies on manual operation. Specifically, after the test report is generated, the operator needs to manually compare multiple data sources such as original records, equipment usage logs, standard method requirements, and sample information to verify the consistency of data, the validity of equipment (such as calibration status and usage permissions), and the logical rationality of the results.

[0005] At this time, although some systems provide basic data export or simple rule prompts, the verification process still requires cross-interface, cross-document, and even cross-system information verification, lacking a unified automated verification mechanism and intelligent judgment capability.

[0006] Therefore, traditional manual verification methods still have significant technical bottlenecks: on the one hand, they rely heavily on human experience and are prone to subjective errors, especially in batch testing tasks where there is high repetition and a large amount of information, making it easy to miss or misjudge; on the other hand, existing systems are difficult to dynamically adapt to different testing items, standards and specifications or customer customization requirements, resulting in rigid verification rules and poor scalability.

[0007] Furthermore, during the manual verification process, there is a lack of an effective automatic alignment mechanism between multi-source heterogeneous data (such as instrument output, manual input, and external standard libraries), which further restricts the verification efficiency and accuracy.

[0008] To address these issues, this application proposes an intelligent verification system and method for test reports, which enables full automation of the test report verification process, configurable rules, and scenario adaptability. Summary of the Invention

[0009] The main objective of this invention is to provide an intelligent verification system and method for test reports, which can realize automated verification of test data, cross-system data linkage, traceability of the verification process, and effective utilization of data, so as to solve the technical problems mentioned in the background art.

[0010] The present invention solves the above-mentioned technical problems by adopting the following technical solutions:

[0011] A smart verification system for test reports, comprising:

[0012] The verification task management module is used to send task information for starting tasks through API calls on a group of lims systems, and save the tasks to the local system after receiving them.

[0013] The prompt word management module is divided into prompt word common part management and detection item management. It is used to supplement and expand the association of anchor keywords containing specified information, and to perform point-to-point precise verification of specific check items, respectively.

[0014] The detection result management module is used to obtain the tasks stored in the local system by the verification task management module, and reconstruct the information value of the detection report according to the usage scenario after the prompt word management module processes the task content;

[0015] The verification task interface module is used to convert the text content in the detection result management module into code structure through verification and display the verification results on the front-end page.

[0016] Preferably, the specific management process for the common part of prompt words in the prompt word management module includes:

[0017] Based on the structural characteristics of industry documents, information is divided into strong identifiers, attribute descriptions, and parameter values. Dedicated extraction logic is set for different categories, and synonymous information features are divided into core feature layer, auxiliary feature layer, and scene feature layer.

[0018] The core feature layer is matched and compared. When the core feature layer is consistent, the auxiliary feature layer is combined with the difference type, including connector difference and suffix difference, and the field correspondence of the scene feature layer to determine whether they are synonymous information.

[0019] Pre-store specified symbol differences and construct a "symbol-semantic" correspondence dictionary. Establish a dedicated mapping relationship for specified expression differences in the detection device model.

[0020] Preferably, while managing the common parts of the prompt words, a logic for handling the omission or replacement of specified connectors is set up to automatically filter specified non-core symbols during the feature extraction stage, retaining only the core character sequence for matching.

[0021] Preferably, the strong identifier class includes unique identifier information such as sample code and equipment number, the attribute description class includes differential attribute descriptions such as model and specifications, and the parameter value class includes parameter values ​​containing units or symbols.

[0022] Preferably, the core feature layer includes irreplaceable specified information, the auxiliary feature layer includes specified modifiers, symbols, connectors and suffixes, and the scene feature layer is set with the association information of the scene to which the document belongs, which is used to assist matching based on the correspondence of scene fields.

[0023] Preferably, the specific management process of the detection item management in the prompt word management module includes: using a set of control interfaces on the browser page to explain and verify the content of the specified detection item, wherein during the verification process, specified special and non-special items of the inspection item are added for comparison with the preloaded file for specified items.

[0024] Preferably, the specific operational process for reconstructing information value according to usage scenarios in the detection result management module includes:

[0025] For detection reports that construct exclusive mapping relationships, the detection reports corresponding to the task content are denoised according to the usage scenario through tasks stored in the local system, eliminating useless information and reducing the burden of subsequent processing.

[0026] The information after noise reduction is annotated with specific content to clarify its purpose;

[0027] Finally, by combining the symbol-semantic mapping relationship in the text content, information data is associated, and logical connections are established between the contents to form a whole.

[0028] Preferably, the specific operation process of the noise reduction process includes:

[0029] The text is split by newline character, and pure white lines are filtered out. Then, dict.fromkeys(lines) is used to remove duplicates while preserving the original order.

[0030] Get the image size using os.path.getsize(), and filter out specified noise data that does not contribute to the core content, such as small document icons and watermarks, to reduce storage and processing pressure.

[0031] Preferably, the specific operation process for information annotation includes:

[0032] Match a set of tags for a specific purpose to the data in the text information, and add functional tags to the matching of text, tables, and images, including:

[0033] Use jieba.analyse to extract weighted text keywords. The higher the weight, the more core the text keywords. Filter the specified viewpoints by long sentences containing core keywords, and then filter the action items by preset action keywords, thereby matching the text fragments with tags that have core viewpoints and action items.

[0034] First, retrieve the specified keywords from the table content as topic tags, and then extract the main content of the main table using the topic tags;

[0035] The final labeled data is structured data with attributes.

[0036] Preferably, the specific operational process for the information data association includes:

[0037] The table's theme keywords are obtained by using `table_keywords = table["theme"].split("、")`. Then, paragraphs containing the table's keywords are filtered from the text. The `if theme in keywords` method is used to check if there are exact matching theme tags for the keywords to ensure accurate association.

[0038] If a perfect match is found, the text, tables, and images related to the keywords in each document will be associated with the same topic, while deduplication and sorting will be performed, ultimately achieving the association of information data related to the text.

[0039] If it is not a perfect match, the text is split into individual words using jieba.lcut(text), and then all keywords are extracted using tfidf_vectorizer.get_feature_names_out(). Then, the total weight of each keyword (importance across all documents) is calculated using tfidf_matrix.sum(axis=0).A[0]. Finally, the keywords are sorted and filtered to remove low-weight (<0.1) main tags to ensure the core of the key.

[0040] Then, the number, detection item and content are extracted from the acquired content. Finally, the DeepSeek-R1-Distill-Qwen-32B tool is used to obtain the final result value, which is then sorted according to keywords and output to achieve the association of information data in related texts.

[0041] The result values ​​are organized according to keywords and output in the following structure: [{"validationItem": "keyword","validationResult": "whether it meets the requirements","reason": "reason for whether it meets the requirements","itemId": "xx","keyWordInfo": [{"keyWord": "inspected unit","page": "corresponding page number"}],"id": "primary key"},...].

[0042] Another invention discloses an intelligent verification method for test reports, implemented based on any of the aforementioned intelligent verification systems for test reports, comprising:

[0043] The lims system calls the verification task management module to obtain task information and receive corresponding tasks.

[0044] After starting the task from the local system, the text is obtained through the detection result management module, and the corresponding prompt words are obtained from the corresponding detection report of the task using the prompt word management module, and the corresponding prompt word association verification is performed.

[0045] Then, the detection result management module is used to reconstruct the information value of the detection reports corresponding to the task content according to the usage scenario;

[0046] Finally, the verification task interface module is used to adjust the text content in the detection result management module into a code structure and display it on the system device front-end page.

[0047] In another aspect, the present invention also discloses a computer device, including a memory, a processor, and a display. The memory stores a computer program for loading the above-mentioned intelligent verification system for test reports. When the computer program is executed by the processor, the processor performs the steps of the above-mentioned intelligent verification method for test reports and finally displays a specified page through the display.

[0048] As can be seen from the above technical solution, the present invention provides an intelligent verification system and method for test reports. Compared with the prior art, the present invention has the following advantages:

[0049] 1. This invention automates the verification process by integrating the verification function, enabling operators to quickly trigger and achieve fully automated verification without the need for manual cross-document and cross-system data comparison. This improves verification efficiency by over 90%, reduces human error, and significantly enhances verification efficiency and accuracy, making it suitable for batch detection tasks.

[0050] 2. This invention combines specific inspection item management prompts and reconstructs the value of inspection report information according to usage scenarios, enabling the construction of a full-dimensional verification system framework covering general scenarios and subdivided fields, thus solving common problems in existing technologies such as multi-source data alignment and dynamic rule adaptation.

[0051] It should be understood that the descriptions in this section are not intended to identify key or essential features of embodiments of the invention, nor are they intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Of course, implementing any product of the invention does not necessarily require achieving all of the advantages described above simultaneously. Attached Figure Description

[0052] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0053] Figure 1 This is a schematic diagram of the overall system flow of the present invention;

[0054] Figure 2 This is a schematic diagram of the verification result output of the present invention;

[0055] Figure 3 This is a schematic diagram of the front-end page display of the present invention. Detailed Implementation

[0056] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] For details in the embodiments, please refer to Figures 1 to 3 .

[0058] like Figure 1 As shown. The intelligent verification system for test reports proposed in this embodiment of the invention includes:

[0059] (a) Verification Task Management Module: On the lims system, a set of API interfaces are called to send task information for starting the task. After receiving the task, this module saves the task to the local system.

[0060] This feature automatically retrieves data from the equipment and sample systems via a pre-defined interface, enabling quick access to related information based on the equipment and sample numbers without requiring manual system switching. This further ensures data consistency and traceability, and enhances the convenience of data retrieval.

[0061] At this point, standardized management of data collection, transmission, verification, and storage can be achieved through unified interface specifications and verification rules. The accumulated verification data can support subsequent data tracking analysis, process optimization, and decision-making, thus further enhancing the value of the data.

[0062] (ii) Prompt word management module: It is divided into prompt word common part management and detection item management, which are used to supplement and expand the association of anchor keywords containing specified information, and to perform point-to-point precise verification of specific inspection items.

[0063] (1) In the embodiment, the common part management of prompt words performs the following operations:

[0064] The core benchmark status of the original record should be clearly defined. The key information contained therein (such as the unique sample code, core detection parameters, etc.) should be used as "anchor keywords" and only used to supplement and expand the association of the anchor information to avoid matching deviations caused by confusion of multi-source information.

[0065] Based on the structural characteristics of documents in the testing industry, information is divided into three categories: "strongly identifiable" (such as sample codes and equipment numbers, which are unique), "attribute descriptions" (such as models and specifications, which are easily differentiated), and "parameter values" (such as test temperature and concentration, which include units or symbols). Dedicated extraction logic is set for each category.

[0066] Synonymous information features are divided into "core feature layer" (irreplaceable key information, such as core values ​​in model number, numerical sequences in code), "auxiliary feature layer" (easily changeable modifiers or symbols, such as connectors like "-", " ", "_", and suffixes like "type" and "specification"), and "scene feature layer" (specific information of the scene to which the document belongs, such as "sample model" in the test report corresponding to "model specification" in the sample information table, and auxiliary matching based on the correspondence of scene fields).

[0067] During matching, the core feature layer is compared first. If the core feature layer matches, the difference type of the auxiliary feature layer (such as connector difference, suffix difference) is combined with the field correspondence of the scene feature layer to determine whether it is synonymous information, rather than using the traditional "full character comparison" or "simple synonym replacement" logic.

[0068] It pre-stores unique symbol differences in the testing industry, constructs a "symbol-semantic" correspondence dictionary, and establishes exclusive mapping relationships for common differences in the expression of testing equipment models, such as "Φ" and "diameter", "℃" and "Celsius", and "±" and "positive and negative". At the same time, it sets up "connector irrelevant" processing logic for common industry-wide cases of connector omission or replacement, such as "XX-100" and "XX 100" and "XX_100", and automatically filters non-core symbols such as "-", " " and "_" during the feature extraction stage, retaining only the core character sequences for matching.

[0069] It should be noted that, due to the characteristics of the testing industry, such as "significant differences between fields and rapid standard updates", this application needs to break through the limitations of traditional verification systems where "fixed rules are difficult to modify" and design a configurable visual configuration rule engine to ensure that even non-technical personnel can add new rules within 5 minutes without modifying the code. This solves the pain point of traditional systems where "adaptation to new fields requires secondary development" and improves system reusability and update efficiency.

[0070] (2) In the embodiment, the following operations are performed in the detection item management:

[0071] Unlike the common content of public prompts, it is used as a supplement to the public prompt module, or for point-to-point precise verification of specific check items.

[0072] Specifically, in the embodiment, a set of control interfaces is used on the browser page to explain and verify the content of the specified detection items. During the verification process, the specified special and non-special items of the inspection items are added to compare the specified items with the preloaded file.

[0073] Specifically, the test items need to be specified, i.e., the specific inspection items; the test item description is an explanation of the specific inspection items.

[0074] Furthermore, the verification logic of the inspection items at this time is not limited to the specific content inspection of a specific item. By adding various special and non-special items in the inspection item design, it only needs to receive the requirement description of comparing this inspection with the specific document, as well as the specific items of the comparison inspection.

[0075] The source of the inspection items must be clearly stated in this item, specifying the documents used for comparison in the inspection logic.

[0076] Finally, this specific check item is generated using the control. At this point, the content of this check item can also be modified according to industry-specific standards.

[0077] (iii) Used to obtain the tasks stored in the local system by the verification task management module, and to reconstruct the information value of the detection report after the prompt word management module processes the task content according to the usage scenario.

[0078] The core of this part of the process is to break the limitations of the original document structure and reconstruct the value of information according to the usage scenario. It is not only about organizing content, but also about making the document have unique practicality (such as decision support, fast retrieval, and cross-scenario adaptation) through information refinement and logical reorganization.

[0079] The process involves noise reduction, information annotation, and data association in sequence to filter out invalid information, label data, and establish connections between content. Noise reduction, information annotation, and data association are sequential processes.

[0080] (1) Noise reduction: It is necessary to remove useless data to reduce the burden of subsequent processing. Specific steps include:

[0081] First, split the text by newline character and filter out pure blank lines (if line.strip()). Then, use dict.fromkeys(lines) to achieve "deduplication while preserving the original order". Get the image size using os.path.getsize() and filter out small document icons, watermarks, etc. that do not contribute to the core content. Filtering reduces storage and processing pressure.

[0082] (2) Information labeling: Label the data to clarify its purpose. Specific steps include:

[0083] Add "functional tags" to text, tables, and images to enable the program to obtain the function and attribute information of the data. For text, use jieba.analyse to extract keywords (with weights; higher weights indicate more core information), then filter key viewpoints through long sentences containing core keywords, and filter action items by specifying "action keywords" (labeling text fragments with "core viewpoints" and "action items"). For tables, first extract keywords from the table content as "topic tags," then extract the main content of the table. For example, if the main tag is "pulse 1," the extracted content will be pulse parameters, values, and pulse form, labeled with "pulse attribute tags" so that the table can be recognized by the program for pulse data analysis. The labeled data is no longer plain text / plain table content, but structured data with attributes.

[0084] Final output:

[0085] Text: {"cleaned": "xxx", "key_points": ["Core viewpoint 1"], "action_items":["XX needs to be completed"]}, Table: {"theme": "Pulse 1", "stats": {"avg": {"arguments": xxx}, "value": {"xxx": xxx},...}}

[0086] (3) Data association: Combine the symbol-semantic mapping relationship in the text content to associate information data, establish logical connections between the contents, and make the content form a whole. The specific steps include:

[0087] First, obtain the theme keywords of the table by table_keywords = table["theme"].split("、"), then filter paragraphs containing the table keywords from the text, and finally associate related texts by combining the symbol-semantic mapping relationship in the text content. At this time, by using all the keywords, associate the text, tables and pictures related to the keywords in each document with the same theme, and simultaneously achieve deduplication and sorting. Use if theme in keywords to determine whether there is a completely matching subject tag for the keyword to ensure the accuracy of association. If it is not completely matched, proceed to the next step, use jieba.lcut(text) to split the text into individual words, then use tfidf_vectorizer.get_feature_names_out() to extract all keywords, then use tfidf_matrix.sum(axis=0).A[0] to calculate the total weight of each keyword (importance across all documents), and finally sort and filter, filtering low weight (<0.1) subject tags to ensure the core of the key. Then, the acquired content is processed using the DeepSeek-R1-Distill-Qwen-32B tool through the logic of 'number + detection item + acquired content' to obtain the final result value. The results are then organized according to keywords into this structure: [{"validationItem": "keyword","validationResult": "whether it conforms","reason": "reason for conformity","itemId": "xx","keyWordInfo": [{"keyWord": "inspected unit","page": "corresponding page number"}],"id": "primary key"},...] Output (e.g.) Figure 2 (As shown)

[0088] In summary, by combining specific inspection item management prompts and reconstructing the value of test report information according to usage scenarios, a full-dimensional verification system framework covering general scenarios and subdivided fields can be constructed. This solves common problems in existing technologies, such as multi-source data alignment and dynamic rule adaptation. Among them, specific inspection item management prompts enable precise implementation in subdivided scenarios. The two work together to break through the efficiency and accuracy bottlenecks of traditional manual verification. The system uses original records as the benchmark, the rule engine as the core, and prompts as verification details. The combination of these elements ensures the accuracy, compliance, and standardization of test reports.

[0089] (iv) Used to adjust and verify the text content in the detection result management module into code structure, and display the verification results on the front-end page, such as Figure 3 As shown.

[0090] In summary, this system, through the automated integration of verification functions, enables operators to quickly trigger and achieve fully automated verification throughout the entire process. It eliminates the need for manual data comparison across documents and systems, improving verification efficiency by over 90%, reducing human error, and significantly enhancing verification efficiency and accuracy. It is suitable for batch detection task scenarios.

[0091] On the other hand, the present invention also discloses an intelligent verification method for test reports, implemented based on the intelligent verification system for test reports in the above embodiments, comprising:

[0092] The lims system calls the verification task management module to obtain task information and receive corresponding tasks.

[0093] After starting the task from the local system, the text is obtained through the detection result management module, and the corresponding prompt words are obtained from the corresponding detection report of the task using the prompt word management module, and the corresponding prompt word association verification is performed.

[0094] Then, the detection result management module is used to reconstruct the information value of the detection reports corresponding to the task content according to the usage scenario;

[0095] Finally, the verification task interface module is used to adjust the text content in the detection result management module into a code structure and display it on the system device front-end page.

[0096] At this time, the device's front-end page can also save the original data before verification and the modified data in real time, and display the comparison information through a visual interface. At the same time, the data is uploaded for data tracking analysis when the device is closed, so as to realize data traceability throughout the verification process, which is convenient for subsequent quality audits and problem tracing.

[0097] In another aspect, the present invention also discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the method described above.

[0098] In another aspect, the present invention also discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the method described above.

[0099] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute any of the intelligent verification methods for test reports in the above embodiments based on the intelligent verification system for test reports in the above embodiments.

[0100] It is understood that the system provided in the embodiments of the present invention corresponds to the method provided in the embodiments of the present invention, and the explanation, examples and beneficial effects of the relevant content can be referred to the corresponding parts of the above methods.

[0101] This application also provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, communication interface, and memory communicate with each other via the communication bus.

[0102] Memory, used to store computer programs;

[0103] When the processor executes the program stored in the memory, it implements the corresponding intelligent verification method for the test report based on the above-mentioned intelligent verification system for test reports.

[0104] The communication bus mentioned in the above-mentioned electronic devices can be a standard bus for interconnecting peripheral components or an extended industrial standard structure bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc.

[0105] The communication interface is used for communication between the aforementioned electronic devices and other devices.

[0106] The memory may include random access memory or non-volatile memory, such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0107] The processors mentioned above can be general-purpose processors, including central processing units, network processors, etc.; they can also be digital signal processors, application-specific integrated circuits, field-programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0108] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, an optical medium, or a semiconductor medium, etc.

[0109] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0110] Furthermore, it should be noted that if any directional indication (such as up, down, left, right, front, back, etc.) is involved in the embodiments of the present invention, the directional indication is only used to explain the relative positional relationship and movement of each component in a specific posture. If the specific posture changes, the directional indication will also change accordingly.

[0111] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the meaning of "and / or" throughout the text includes three parallel solutions; for example, "A and / or B" includes solution A, solution B, or a solution where both A and B are satisfied simultaneously. Furthermore, in the embodiments of this invention, "multiple" refers to two or more. Moreover, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

Claims

1. An intelligent verification system for test reports, characterized in that, include: The verification task management module is used to send task information for starting tasks through API calls on a group of lims systems, and save the tasks to the local system after receiving them. The prompt word management module is divided into prompt word common part management and detection item management. It is used to supplement and expand the association of anchor keywords containing specified information, and to perform point-to-point precise verification of specific check items, respectively. The detection result management module is used to obtain the tasks stored in the local system by the verification task management module, and reconstruct the information value of the detection report according to the usage scenario after the prompt word management module processes the task content; The verification task interface module is used to convert the text content in the detection result management module into code structure through verification and display the verification results on the front-end page. The specific management process for the common part of prompt words in the prompt word management module includes: Based on the structural characteristics of industry documents, information is divided into strong identifiers, attribute descriptions, and parameter values. Dedicated extraction logic is set for different categories, and synonymous information features are divided into core feature layer, auxiliary feature layer, and scene feature layer. The core feature layer is matched and compared. When the core feature layer is consistent, the auxiliary feature layer is combined with the difference type, including connector difference and suffix difference, and the field correspondence of the scene feature layer to determine whether they are synonymous information. Pre-store specified symbol difference combinations, construct a "symbol-semantic" correspondence dictionary, and establish a dedicated mapping relationship for specified expression differences in the detection device model; Meanwhile, for cases where specified connectors are omitted or replaced, connector irrelevant processing logic is set up to automatically filter specified non-core symbols during the feature extraction stage, retaining only the core character sequence for matching; The core feature layer includes irreplaceable specified information such as core numerical values ​​in the model number and numerical sequences in the code. The auxiliary feature layer includes specified modifiers, symbols, connectors, and suffixes. The scene feature layer sets the association information of the scene to which the document belongs, which is used to assist in matching based on the correspondence of scene fields.

2. The intelligent verification system for test reports as described in claim 1, characterized in that, The strong identifier class includes unique identifier information such as sample code and equipment number; the attribute description class includes differential attribute descriptions such as model and specifications; and the parameter value class includes parameter values ​​containing units or symbols.

3. The intelligent verification system for test reports as described in claim 1, characterized in that, The specific management process of the detection item management in the prompt word management module includes: using a set of control interfaces on the browser page to explain and verify the content of the specified detection item, wherein the verification process of the detection item is carried out by adding specified special and non-special items of the inspection item, which are used to compare the specified items with the preloaded file.

4. The intelligent verification system for test reports as described in claim 3, characterized in that, The specific operational procedures for reconstructing information value according to usage scenarios in the detection result management module include: For detection reports that construct exclusive mapping relationships, the detection reports corresponding to the task content are denoised according to the usage scenario through tasks stored in the local system, eliminating useless information and reducing the burden of subsequent processing. The information after noise reduction is annotated with specific content to clarify its purpose; Finally, by combining the symbol-semantic mapping relationship in the text content, information data is associated, and logical connections are established between the contents to form a whole.

5. The intelligent verification system for test reports as described in claim 4, characterized in that, The specific operation process of the noise reduction process includes: The text is split by newline character, and pure white lines are filtered out. Then, dict.fromkeys(lines) is used to remove duplicates while preserving the original order. Get the image size using os.path.getsize(), and filter out specified noise data that does not contribute to the core content, such as small document icons and watermarks, to reduce storage and processing pressure.

6. The intelligent verification system for test reports as described in claim 4, characterized in that, The specific operation process for annotating the noise-reduced information in the detection result management module includes: Match a set of tags for a specific purpose to the data in the text information, and add functional tags to the matching of text, tables, and images, including: Use jieba.analyse to extract weighted text keywords, filter for specific viewpoints by long sentences containing core keywords, and then filter for action items by preset action keywords, thereby matching text fragments with tags that contain core viewpoints and action items; First, retrieve the specified keywords from the table content as topic tags, and then extract the main content of the main table using the topic tags; The final labeled data is structured data with attributes.

7. The intelligent verification system for test reports as described in claim 4, characterized in that, The specific operational process for linking the information data includes: The table's theme keywords are obtained by using `table_keywords = table["theme"].split("、")`. Then, paragraphs containing the table's keywords are filtered from the text. The `if theme in keywords` method is used to check if there are exact matching theme tags for the keywords to ensure accurate association. If a perfect match is found, the text, tables, and images related to the keywords in each document will be associated with the same topic, while deduplication and sorting will be performed, ultimately achieving the association of information data related to the text. If it is not a complete match, the text is split into individual words using jieba.lcut(text), and then all keywords are extracted using tfidf_vectorizer.get_feature_names_out(). Then, the total weight of each keyword is calculated using tfidf_matrix.sum(axis=0).A[0]. Finally, the text is sorted and filtered to remove low-weight main tags to ensure the core of the key. Then, the number, detection item, and content are extracted from the acquired content. Finally, the DeepSeek-R1-Distill-Qwen-32B tool is used to obtain the final result value, which is then sorted according to keywords and output to achieve the association of information data in related texts.

8. A method for intelligent verification of test reports, implemented based on the intelligent verification system for test reports according to any one of claims 1-7, characterized in that, include: The lims system calls the verification task management module to obtain task information and receive corresponding tasks. After starting the task from the local system, the text is obtained through the detection result management module, and the corresponding prompt words are obtained from the corresponding detection report of the task using the prompt word management module, and the corresponding prompt word association verification is performed. Then, the detection result management module is used to reconstruct the information value of the detection reports corresponding to the task content according to the usage scenario; Finally, the verification task interface module is used to adjust the text content in the detection result management module into a code structure and display it on the system device front-end page.

Images