Test item determination method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides a test item determination method and device, electronic equipment and a storage medium. The test item determination method comprises the following steps: acquiring at least one reference document associated with a signal test of a target signal; and identifying the at least one reference document by adopting a language identification model, and extracting test item information of a test item associated with the signal test of the target signal from the at least one reference document.

Description

Technical Field

[0001] This invention relates to the field of signal measurement technology, and in particular to a method, apparatus, electronic device, and storage medium for determining test items. Background Technology

[0002] With the rapid development of information technology, signal testing plays an increasingly important role in fields such as communications, electronics, and aerospace. Determining appropriate test items is crucial for ensuring test quality and efficiency during signal testing. For example, the test items differ depending on the transmission protocol's signal compliance requirements. Appropriate test items are necessary to ensure that the signal meets the requirements of the transmission protocol. Traditional methods for determining test items mainly rely on manually extracting relevant information from various reference documents. This approach is not only time-consuming and labor-intensive but also prone to omissions and errors. Summary of the Invention

[0003] In view of this, embodiments of the present disclosure provide a method, apparatus, electronic device, and storage medium for determining test items.

[0004] According to a first aspect of the embodiments of this disclosure, a method for determining test items is provided, the method comprising:

[0005] Obtain at least one reference document associated with the signal testing of the target signal;

[0006] A language recognition model is used to identify the at least one reference document, and test item information of the target signal associated with the test item is extracted from the at least one reference document.

[0007] In some embodiments, the method further includes: preprocessing the at least one reference document;

[0008] The step of using a language recognition model to identify the at least one reference document includes: using the language recognition model to identify the at least one reference document that has undergone preprocessing;

[0009] The preprocessing of the at least one reference document includes at least one of the following:

[0010] The reference document is parsed using a parsing library corresponding to its format to obtain a reference document in a predetermined format;

[0011] The reference document is converted into a text sequence of reference documents;

[0012] The reference document is segmented using a predetermined segmentation method, wherein the sub-blocks of the reference document obtained by the segmentation are respectively recognized by the language recognition model.

[0013] In some embodiments, segmenting the reference document using a predetermined segmentation method includes at least one of the following:

[0014] Divide according to the chapters of the referenced document;

[0015] Divide according to the title of the referenced document;

[0016] Divide according to the paragraphs of the reference document;

[0017] The reference document is segmented according to a predetermined data volume;

[0018] The reference document is segmented according to its semantics.

[0019] In some embodiments, the method further includes at least one of the following:

[0020] Perform a deduplication operation on the test item information of the test items;

[0021] Merge test project information for the same test project;

[0022] Mark test items that do not meet the preset conditions.

[0023] In some embodiments, the deduplication operation on the test item information of the test item includes at least one of the following:

[0024] Deduplication is performed based on the vector cosine similarity between multiple test items of the same test item;

[0025] A comparison model is used to deduplicate test item information for the same test item.

[0026] In some embodiments, the merging of test project information for the same test project includes:

[0027] Retain the first test item information with the highest priority among multiple test item information for the same test item;

[0028] The second test item information from the plurality of test item information is merged into the first test item information to obtain the merged test item information;

[0029] The priority of each test item information among the plurality of test item information is determined based on at least one of the following: the type of reference document corresponding to each test item information; the amount of information contained in each test item information.

[0030] In some embodiments, the method further includes: establishing a tree structure containing multiple test items, wherein each node of the tree structure corresponds to at least one test item.

[0031] In some embodiments, establishing a tree structure containing multiple test items includes at least one of the following:

[0032] Based on the hierarchical relationship of the test items in the corresponding reference documents, determine the level of the test items in the tree structure;

[0033] The level of the test item in the tree structure is determined based on the judgment model;

[0034] The hierarchy of the test items in the tree structure is determined based on predetermined rules;

[0035] At least one root node is set based on the standard and / or the function of the target signal association.

[0036] In some embodiments, the method further includes:

[0037] Each node of the tree structure is assigned a corresponding identifier, and the tree structure is characterized by the connection relationship information of the identifiers.

[0038] In some embodiments, the test item information is used to indicate at least one of the following:

[0039] The name of the test item;

[0040] The settings for the test items;

[0041] The test parameters of the test items;

[0042] The judgment criteria for the test items;

[0043] Source information of the test items;

[0044] The description information of the test items.

[0045] According to a second aspect of the present disclosure, a test item determination apparatus is provided, characterized in that the apparatus includes a processing module, wherein the processing module is configured to:

[0046] Obtain at least one reference document associated with the signal testing of the target signal;

[0047] A language recognition model is used to identify the at least one reference document, and test item information of the target signal associated with the test item is extracted from the at least one reference document.

[0048] In some embodiments, the processing module is further configured to: preprocess the at least one reference document;

[0049] The processing module is specifically used to: identify the at least one reference document that has undergone preprocessing using the language recognition model;

[0050] The processing module is specifically used for at least one of the following:

[0051] The reference document is parsed using a parsing library corresponding to its format to obtain a reference document in a predetermined format;

[0052] The reference document is converted into a text sequence of reference documents;

[0053] The reference document is segmented using a predetermined segmentation method, wherein the sub-blocks of the reference document obtained by the segmentation are respectively recognized by the language recognition model.

[0054] In some embodiments, the processing module is specifically used for at least one of the following:

[0055] Divide according to the chapters of the referenced document;

[0056] Divide according to the title of the referenced document;

[0057] Divide according to the paragraphs of the reference document;

[0058] The reference document is segmented according to a predetermined data volume;

[0059] The reference document is segmented according to its semantics.

[0060] In some embodiments, the processing module is further configured to perform at least one of the following:

[0061] Perform a deduplication operation on the test item information of the test items;

[0062] Merge test project information for the same test project;

[0063] Mark test items that do not meet the preset conditions.

[0064] In some embodiments, the processing module is specifically used for at least one of the following:

[0065] Deduplication is performed based on the vector cosine similarity between multiple test items of the same test item;

[0066] A comparison model is used to deduplicate test item information for the same test item.

[0067] In some embodiments, the processing module is specifically used for:

[0068] Retain the first test item information with the highest priority among multiple test item information for the same test item;

[0069] The second test item information from the plurality of test item information is merged into the first test item information to obtain the merged test item information;

[0070] The priority of each test item information among the plurality of test item information is determined based on at least one of the following: the type of reference document corresponding to each test item information; the amount of information contained in each test item information.

[0071] In some embodiments, the processing module is further configured to: establish a tree structure containing multiple test items, wherein each node of the tree structure corresponds to at least one test item.

[0072] In some embodiments, the processing module is specifically used for at least one of the following:

[0073] Based on the hierarchical relationship of the test items in the corresponding reference documents, determine the level of the test items in the tree structure;

[0074] The level of the test item in the tree structure is determined based on the judgment model;

[0075] The hierarchy of the test items in the tree structure is determined based on predetermined rules;

[0076] At least one root node is set based on the standard and / or the function of the target signal association.

[0077] In some embodiments, the processing module is further configured to:

[0078] Each node of the tree structure is assigned a corresponding identifier, and the tree structure is characterized by the connection relationship information of the identifiers.

[0079] In some embodiments, the test item information is used to indicate at least one of the following:

[0080] The name of the test item;

[0081] The settings for the test items;

[0082] The test parameters of the test items;

[0083] The judgment criteria for the test items;

[0084] Source information of the test items;

[0085] The description information of the test items.

[0086] According to a third aspect of the present disclosure, an electronic device is provided, including a processor, a memory, and an executable program stored in the memory and operable by the processor, characterized in that the processor executes the steps of the test item determination method as described in the first aspect when running the executable program.

[0087] According to a fourth aspect of the present disclosure, a storage medium is provided that stores an executable program thereon, characterized in that the executable program, when executed by a processor, implements the steps of the test item determination method as described in the first aspect.

[0088] The test item determination method provided in this disclosure automatically extracts test item information from reference documents using a language recognition model. Since the language recognition model is efficient and accurate, it can improve the efficiency and accuracy of test item information determination, reduce the human resource cost of manually determining test items, and reduce errors and omissions that may result from manual determination of test items. Attached Figure Description

[0089] Figure 1 This is a flowchart illustrating a method for determining test items according to an exemplary embodiment;

[0090] Figure 2 This is a schematic diagram illustrating the composition of an interactive system according to an exemplary embodiment;

[0091] Figure 3 This is a flowchart illustrating another method for determining test items according to an exemplary embodiment;

[0092] Figure 4 This is a schematic diagram of a test item determination device according to an exemplary embodiment;

[0093] Figure 5 This is a schematic diagram of an electronic device structure according to an exemplary embodiment. Detailed Implementation

[0094] To make the technical solution and beneficial effects of the present invention more apparent and understandable, a detailed description is provided below by listing specific embodiments. The accompanying drawings are not necessarily drawn to scale, and local features may be enlarged or reduced to more clearly show the details of the local features; unless otherwise defined, the technical and scientific terms used herein have the same meanings as those in the technical field to which this application pertains.

[0095] This disclosure is not exhaustive, but merely illustrative of some embodiments, and is not intended to limit the scope of protection of this disclosure. Unless otherwise specified, each step in a particular embodiment can be implemented as an independent embodiment, and the steps can be arbitrarily combined. For example, a solution after removing some steps in a particular embodiment can also be implemented as an independent embodiment, and the order of the steps in a particular embodiment can be arbitrarily interchanged. Furthermore, the optional implementation methods in a particular embodiment can be arbitrarily combined; moreover, the embodiments can be arbitrarily combined, for example, some or all steps of different embodiments can be arbitrarily combined, and a particular embodiment can be arbitrarily combined with the optional implementation methods of other embodiments.

[0096] In each of the disclosed embodiments, unless otherwise specified or in case of logical conflict, the terminology and / or descriptions of the embodiments are consistent and can be referenced by each other. Technical features in different embodiments can be combined to form new embodiments based on their inherent logical relationships.

[0097] The terminology used in the embodiments of this disclosure is for the purpose of describing particular embodiments only and is not intended to limit the scope of this disclosure.

[0098] In this embodiment of the disclosure, unless otherwise stated, elements expressed in the singular form, such as "a," "an," "the," "the," "the," "the," "the," "the," "this," etc., can mean "one and only one," or "one or more," "at least one," etc. For example, when using articles such as "a," "an," "the," etc. in translation, the noun following the article can be understood as either a singular expression or a plural expression.

[0099] In the embodiments disclosed herein, "multiple" refers to two or more.

[0100] In some embodiments, the terms “at least one of”, “one or more”, “a plurality of”, “multiple”, etc., may be used interchangeably.

[0101] In some embodiments, the notation "at least one of A and B", "A and / or B", "A in one case, B in another", "A in one case, B in another", etc., may include the following technical solutions depending on the situation: in some embodiments, A (A is executed regardless of B); in some embodiments, B (B is executed regardless of A); in some embodiments, execution is selected from A and B (A and B are selectively executed); in some embodiments, A and B (both A and B are executed). The same applies when there are more branches such as A, B, C, etc.

[0102] In some embodiments, the notation "A or B" may include the following technical solutions, depending on the situation: in some embodiments, A (execution of A regardless of B); in some embodiments, B (execution of B regardless of A); in some embodiments, execution is selected from A and B (A and B are selectively executed). The same applies when there are more branches such as A, B, C, etc.

[0103] The prefixes "first," "second," etc., used in the embodiments of this disclosure are merely for distinguishing different descriptive objects and do not impose restrictions on the position, order, priority, value, or content of the descriptive objects. The description of the descriptive objects should be found in the claims or the context of the embodiments, and the use of prefixes should not constitute unnecessary restrictions. For example, if the descriptive object is a "field," the ordinal numbers preceding "field" in "first field" and "second field" do not restrict the position or order of the "fields." "First" and "second" do not restrict whether the "fields" they modify are in the same message, nor do they restrict the order of "first field" and "second field." Similarly, if the descriptive object is a "level," the ordinal numbers preceding "level" in "first level" and "second level" do not restrict the priority between "levels." Furthermore, the value of the descriptive object is not limited by ordinal numbers and can be one or more. For example, in "first device," the value of "device" can be one or more. Furthermore, the objects modified by different prefixes can be the same or different. For example, if the object being described is "device", then "first device" and "second device" can be the same device or different devices, and their types can be the same or different. Similarly, if the object being described is "information", then "first information" and "second information" can be the same information or different information, and their content can be the same or different.

[0104] In some embodiments, “including A,” “containing A,” “for indicating A,” and “carrying A” can be interpreted as directly carrying A or indirectly indicating A.

[0105] In some embodiments, terms such as “…”, “determine…”, “in the case of…”, “when…”, “when…”, “if…”, etc. can be used interchangeably.

[0106] In some embodiments, the terms “greater than,” “greater than or equal to,” “not less than,” “more than,” “more than or equal to,” “not less than,” “higher than,” “higher than or equal to,” “not lower than,” and “above” can be used interchangeably, as can the terms “less than,” “less than or equal to,” “not greater than,” “less than,” “less than or equal to,” “not more than,” “lower than,” “lower than or equal to,” “not higher than,” and “below”.

[0107] In some embodiments, devices, etc., can be interpreted as physical or virtual, and their names are not limited to the names recorded in the embodiments. Terms such as “device”, “equipment”, “circuit”, “network element”, “node”, “function”, “unit”, “section”, “system”, “network”, “chip”, “chip system”, “entity”, and “subject” can be used interchangeably.

[0108] Furthermore, each element, each row, or each column in the table of this disclosure can be implemented as an independent embodiment, and any combination of any element, any row, or any column can also be implemented as an independent embodiment.

[0109] Signal measurement equipment, such as oscilloscopes, plays a crucial role in verifying whether different high-speed interfaces (such as USB, PCIe, HDMI, DisplayPort, Ethernet, DDR, etc.) conform to physical layer standards. Conformance testing ensures interoperability between devices from different manufacturers and is a necessary step for product certification and market entry.

[0110] The test item information required to define the test items associated with these consistency tests and other signal tests is scattered across different documents, for example:

[0111] 1. Official Standard Specifications: These are published by standards organizations (such as USB-IF, PCI-SIG, VESA, IEEE) and are usually hundreds of pages long PDF documents containing detailed test procedures, parameter definitions, templates (masks), and limits.

[0112] 2. Internal test documentation: accumulated test plans, test cases, test procedures, historical test data and experience summaries, which may be in Word, Excel, internal wiki pages or database formats.

[0113] 3. External test documents: Product manuals, datasheets, application notes, white papers, public test reports, etc. published by third parties, used to understand the test coverage, performance indicators and test methods of third-party products, in various formats.

[0114] Currently, the primary method for obtaining signal testing-related test project information relies on manual operation. Test engineers need to invest a significant amount of time manually reading documents from different sources and in different formats. For example:

[0115] 1. Manually extract key information, including the test item name (e.g., “Differential Impedance Test”, “Eye Diagram Mask Test”, “Jitter Measurement – ​​TJ, RJ, DJ”), test conditions / settings (e.g., signal rate, code pattern, oscilloscope bandwidth / sampling rate settings), specific test parameters (e.g., impedance range, eye height / eye width, jitter component values), and pass / fail criteria (limit values ​​or mask).

[0116] 2. Manual comparison and integration: Information extracted from different documents will be compared to identify duplicate or similar test items (even if the names are slightly different, such as "Eye Mask Test" and "Eye Diagram Compliance"), merge information, and resolve conflicts (such as different documents having different limits for the same parameter).

[0117] 3. Manual organization: Organize the integrated test items into a list or simple category structure, and integrate them into a file with a predetermined format.

[0118] 4. Manual update: When the standard is updated or new competitor information is discovered, the above process needs to be repeated, which is a huge workload and can easily introduce new errors.

[0119] There are still many shortcomings in manually determining test item information:

[0120] Extremely time-consuming and costly: Compiling conformance test items for a complex standard (such as PCIe 6.0) can take senior engineers weeks or even months. High human resource costs.

[0121] Highly subjective and prone to errors: Interpretations of standards, focus of information extraction, and judgment of similar test items can all vary from person to person, leading to inconsistent results. Omission of key test items may result in product certification failure or missed market opportunities. Incorrect parameter transcription may lead to incorrect test settings.

[0122] Difficulty in effectively utilizing competitor information: Quickly and accurately extracting valuable test item information from a large number of competitor documents and conducting systematic benchmarking analysis is a huge challenge for humans, often resulting in only point-based and lagging analysis.

[0123] Lack of dynamic updating and maintenance mechanisms: Information on standards and third-party products is constantly changing, making it very difficult to manually maintain the list of test items, which can easily lead to outdated information.

[0124] Information silos and knowledge loss: Manually organized results are often stored on personal computers or in scattered files, making it difficult to form a structured, shareable, and reusable knowledge base, and there is a risk of knowledge loss due to personnel changes.

[0125] Poor traceability and difficulty in verification: When there are doubts about the accuracy or source of a certain integrated test item, it is necessary to review a large number of original documents for verification, which is inefficient.

[0126] Therefore, there is an urgent need for a test item determination method that can automatically extract test item information from multiple reference documents and effectively organize and optimize it.

[0127] This disclosure provides a method for determining test items, such as... Figure 1 As shown, the method includes:

[0128] Step 101: Obtain at least one reference document associated with the signal test of the target signal;

[0129] Step 102: Use a language recognition model to identify the at least one reference document, and extract the test item information of the test item associated with the signal test of the target signal from the at least one reference document.

[0130] In this embodiment, the target signal can be any type of signal that needs to be tested. For example, the target signal may include signals that require compliance testing and / or integrity testing. For example, the target signal may be a signal from a high-speed interface or transmission protocol (such as USB, PCIe, HDMI, DisplayPort, Ethernet, DDR, etc.).

[0131] The target signal can include multiple signals from the same transmission protocol or high-speed interface. For example, the target signal can be the clock signal and the data signal in PCIe.

[0132] Reference documents can be various documents related to the testing of the target signal, such as test standard documents, test specification documents, test report documents, product manual documents, etc.

[0133] Specifically, the process of obtaining at least one reference document associated with the signal testing of the target signal can be achieved in various ways. For example, reference documents related to the target signal can be obtained through database retrieval; reference documents related to the target signal can also be obtained through web search engines; or reference documents can be obtained through manual upload by the user. The obtained reference documents can be in various formats, such as PDF, Word, HTML, TXT, XLSX, CSV, etc., and are not limited here.

[0134] In the process of identifying reference documents using language recognition models, various language recognition models from natural language processing techniques can be used, such as Large Language Models (LLMs), deep learning-based language models, and / or rule-based language models. These language recognition models can understand the document content and identify information relevant to the target signal test.

[0135] In the process of extracting test item information related to the signal test of the target signal from the reference document, the language recognition model can analyze the content of the reference document, identify the test items related to the target signal test, and extract relevant information about these test items. Test item information may include the name of the test item, test parameters, test conditions, test methods, etc.

[0136] The speech recognition model can output test item information based on user-preset or default requirements. For example, a predefined Prompt template can be used. The speech recognition model recognizes a reference document and returns a JSON response based on the Prompt template. Here, the Prompt template is a structured user request used to guide the speech recognition model in generating expected output. The JSON response can be a standardized output format based on the Prompt template to facilitate program parsing and subsequent processing.

[0137] In some embodiments, the test item information is used to indicate at least one of the following: the name of the test item; the settings of the test item; the test parameters of the test item; the judgment conditions of the test item; the source information of the test item; and the description information of the test item.

[0138] Test item information can be information related to the test item extracted from reference documents, used to guide the execution and evaluation of the test item. Test item information can also be information added during the processing and identification of reference documents (e.g., identifiers set for each test item).

[0139] Source information for test items can indicate their origin, such as which reference document they came from, the version of the reference document, the publication date of the reference document, and their specific location within the reference document. Source information helps trace the origin of test items and facilitates access to the original document when needed.

[0140] The name of a test item can be used to identify the test item, such as "differential impedance test," "eye diagram template test," or "jitter measurement." The name of a test item can clearly express the purpose and content of the test item.

[0141] Test setup settings can be used to indicate the configuration information for test items, such as the test environment configuration, test equipment configuration (e.g., signal rate, code pattern, oscilloscope bandwidth and / or sampling rate settings), and test software configuration. These settings help ensure the consistency of the test environment and improve the repeatability of test results.

[0142] Test parameters for a test item can indicate the specific parameters to be measured, such as impedance range, eye height / eye width, and jitter component values. Test parameters should be clearly defined to facilitate testing by personnel.

[0143] Judgment criteria for test items can be used to indicate the evaluation standards for test items, such as jitter should be less than a certain limit, and the signal waveform should conform to the eye diagram template. Judgment criteria help to objectively evaluate test results and determine whether the test passes or fails.

[0144] The descriptive information of a test item can be used to explain the test purpose, object, and scenario. For example, testing the pulse characteristics of a clock signal. This descriptive information helps testers gain an initial understanding of the test item.

[0145] In practical applications, test project information may include one or more of the above-mentioned elements, depending on the information provided in the reference documentation and the requirements of the test project. Extracting complete test project information provides comprehensive guidance for test project execution, improving testing efficiency and accuracy.

[0146] In one possible implementation, the test item information can be used to enable signal measurement equipment such as oscilloscopes to perform measurements on the target signal.

[0147] In one possible implementation, each reference document can be assigned a unique identifier (ID). Reference documents can then be accessed based on this identifier.

[0148] In practical applications, an interactive system can be established for testing projects. Users can upload reference documents for different target signals in advance through the interactive system's interface. These reference documents can be stored on servers, cloud storage, or other locations. The interactive system can record file information for the reference documents (such as ID, name, type, source marker, upload time, and storage path). Users can manage the reference documents through the interactive system, including previewing and deleting them. When users need to determine the test project information for a target signal, they can select the reference document associated with the target signal through the interactive interface and trigger a language recognition model to recognize the selected reference document and identify the test project information within it using trigger commands (such as button presses on the interactive interface).

[0149] The method in this embodiment can be applied to various signal testing scenarios, such as signal testing of electronic products, signal testing of communication equipment, and signal testing of medical equipment.

[0150] In one possible implementation, the test items can be used by measurement equipment such as an oscilloscope to measure the target signal. For example, the oscilloscope can perform measurements based on the test item information and compare the measurement results with the judgment criteria of the test items to determine whether the target signal meets the requirements.

[0151] By using a language recognition model to automatically extract test item information from reference documents, the efficiency and accuracy of test item information determination can be improved due to the high efficiency and accuracy of language recognition models. This reduces the human resource costs associated with manual test item determination and minimizes potential errors and omissions.

[0152] In some embodiments, before inputting the reference document into the language recognition model, the method further includes: preprocessing the at least one reference document.

[0153] Step 102 may include: using the language recognition model to identify the at least one reference document that has undergone preprocessing.

[0154] The preprocessing of the at least one reference document includes at least one of the following: parsing the reference document using a parsing library corresponding to the format of the reference document to obtain a reference document in a predetermined format; converting the reference document into a text sequence reference document; and segmenting the reference document using a predetermined segmentation method, wherein the sub-blocks of the reference document obtained by the segmentation are respectively used for recognition by the language recognition model.

[0155] In this embodiment, the purpose of preprocessing the reference documents is to convert reference documents of different formats and structures into a format that the language recognition model can effectively process, thereby improving the recognition efficiency and accuracy of the language recognition model.

[0156] When parsing reference documents using a parsing library corresponding to the document's format, the appropriate parsing library can be selected based on the document's format. These libraries can parse reference documents of different formats into documents of a predetermined format, such as uniformly converting them into structured data in JSON or XML format for subsequent processing.

[0157] Examples: For PDF reference documents: This requires processing both text-based PDFs and image-based PDFs (the latter requires Optical Character Recognition (OCR)). A PDF parsing library can extract text content, font information, and certain layout / structure information (such as titles and lists). For Word reference documents: A Word parsing library can convert DOCX files to HTML or plain text, preserving titles, lists, tables, and other structures relatively well. For Excel reference documents: An Excel parsing library can read workbooks, worksheets, and cell contents, converting them into JSON arrays or objects while retaining row and column structure. For CSV reference documents: A CSV parsing library can parse CSV files into JSON object arrays.

[0158] Here, text sequences refer to natural language text represented as character sequences. Text sequences follow a standard format, making them easy for language recognition models to parse. Furthermore, they have a smaller data volume, reducing the load on language recognition models and thus improving their recognition efficiency.

[0159] In the process of converting a reference document into a text sequence, the text content can be extracted from the reference document to form a continuous text sequence. This removes non-text content such as formatting information and images from the document, retaining only the text information relevant to the test item, which facilitates processing by the language recognition model.

[0160] In practical applications, during the process of converting a reference document into a text sequence, structural information (such as headings, hierarchy, and table labels) can be preserved or extracted. Irrelevant headers, footers, formatting characters, and other noise can be removed. This improves the accuracy of the language recognition model in recognizing test item information.

[0161] Language recognition models typically have limitations on the length of the text they can recognize. Therefore, reference documents can be segmented to meet the requirements of the language recognition model. By using a predetermined segmentation method, longer reference documents can be divided into multiple smaller parts, making them easier for the language recognition model to process. The segmentation method can be selected according to actual needs, such as segmentation by fixed size, segmentation by chapter, or segmentation by paragraph.

[0162] In one possible implementation, parsing a reference document may include parsing the structure of the reference document and determining its structural information.

[0163] After preprocessing, the language recognition model can process the reference document more efficiently and extract the test item information of the target signal. The preprocessing process can be performed using one or more methods depending on the actual situation to achieve the best preprocessing results.

[0164] In some embodiments, segmenting the reference document using a predetermined segmentation method includes at least one of the following: segmenting by chapters of the reference document; segmenting by titles of the reference document; segmenting by paragraphs of the reference document; segmenting the reference document by a predetermined amount of data; or segmenting the reference document by its semantics.

[0165] When segmenting a reference document by chapter, you can first identify the chapter markers (such as Chapter 1, Chapter 2, etc.) in the reference document and divide the document into multiple parts by chapter. Segmenting a reference document by chapter can be applied to reference documents with a clear chapter structure, such as standard documents and specification documents.

[0166] When segmenting a reference document based on its headings, the system can identify the headings (such as first-level headings, second-level headings, etc.) within the document and divide it into multiple sections according to the headings. This method of segmenting reference documents by headings is suitable for documents with a clear hierarchical heading structure, such as technical reports and product manuals.

[0167] In one possible implementation, the segmentation by chapters and / or headings of the reference document can be based on the reference document structure information determined when parsing the reference document.

[0168] When segmenting a reference document based on paragraphs, it can identify paragraph separators (such as line breaks and blank lines) and divide the document into multiple parts by paragraphs. This segmentation method is applicable to various types of reference documents and is a universal segmentation method.

[0169] When segmenting a reference document based on a predetermined data volume, a data volume threshold (such as the number of tokens, characters, or words) can be set, and segmentation occurs when the threshold is reached. For example, each segmentation unit can be set to 1000-2000 tokens, and the document can be segmented according to this data volume. Segmenting a reference document based on a predetermined data volume is suitable for reference documents with a large amount of data, ensuring that the data volume of each segment is moderate and easy for the language recognition model to process.

[0170] When segmenting a reference document based on semantics, semantic analysis techniques can be used to identify semantic units (such as topics, themes, etc.) within the document, dividing it into multiple parts according to these semantic units. This method requires relatively complex semantic analysis techniques, but it can yield semantically coherent segmentation results, which is beneficial for language recognition models to understand the document content.

[0171] In one possible implementation, the original location information of the sub-blocks obtained by segmenting the reference document within the reference document can be recorded, such as the start and end positions (including page numbers, paragraph numbers, etc.) within the reference document. This original location information can be used to determine the source of test item information identified from the sub-block records.

[0172] In practical applications, one or more segmentation methods can be selected and combined according to the characteristics of the reference document and actual needs to achieve the best segmentation effect.

[0173] On the one hand, using multiple segmentation methods to segment a reference document can improve the flexibility of document segmentation and make it suitable for different segmentation scenarios. On the other hand, segmenting a reference document can divide a large document into multiple smaller parts, making it easier for the language recognition model to process and improving processing efficiency and accuracy.

[0174] After the language recognition model identifies the test item information, the test item information can be processed in the following ways to achieve information integration.

[0175] In some embodiments, the method further includes at least one of the following: performing a deduplication operation on the test item information of the test items; merging the test item information of the same test item; and marking test item information that does not meet preset conditions.

[0176] A speech recognition model can identify one or more test item information entries for the same test item from a single reference file. These entries can be identical or different. Similarly, a speech recognition model can identify one or more test item information entries for the same test item from multiple reference files. These entries can also be identical or different. Therefore, after identifying multiple test item information entries for the same test item, deduplication and merging operations can be performed to reduce duplicate test item information and obtain more complete test item information.

[0177] In one possible implementation, test item information for the same test item can be determined based on the name of the test item in the test item information. For example, preliminary screening can be performed based on the name of the test item (such as using edit distance or Jaccard similarity) to find candidate test item information that may be duplicated, and then the candidate test item information can be deduplicated.

[0178] During the deduplication process, duplicate test item information within the same test item can be identified and removed. When extracting test item information from multiple reference documents, duplication may occur, such as different reference documents containing the same test item information. Deduplication prevents duplicate test item information from interfering with subsequent processing. For example, if the test item information extracted from reference document 1 contains test parameters for jitter testing of the target signal, and the test item information extracted from reference document 2 also contains test parameters for jitter testing of the target signal, then the test item information extracted from document 2 can be discarded. Deduplication reduces the redundancy of test item information, maintaining its simplicity.

[0179] Different reference documents may contain different aspects of the same test item. Test item information from different reference documents that belong to the same test item can be merged. This merging operation yields more complete test item information. For example, if the test item information extracted from reference document 1 contains test parameters for jitter testing of the target signal, and the test item information extracted from reference document 2 contains judgment conditions for jitter testing of the target signal, then the test item information extracted from reference document 1 and reference document 2 can be merged to obtain the test parameters and judgment conditions for jitter testing. This merging operation improves the completeness of the test item information.

[0180] In practical applications, when merging test item names, the most standardized or commonly used names can be selected. For test item descriptions, the most detailed descriptions can be merged or selected. For test parameters or decision conditions (such as limits), if the parameters or conditions differ from different sources, all can be recorded, displayed side-by-side, and their sources labeled. (For example, test parameter: voltage; limit: 800mV (Source: Reference Document A), 750mV (Source: Reference Document B)). For test item source information, the indexes of all original sources can be aggregated into the merged entry.

[0181] Since the test item information is extracted from different reference documents, there may be instances where the extracted information does not meet preset conditions. These instances may include conflicts between test item information, failure to meet information completeness requirements (such as the requirement to include test parameters and decision conditions), or failure to meet information accuracy requirements (such as test parameters needing to be within a reasonable range). For example, two sets of test item information might provide contradictory values ​​for the same test item's parameters. Test item information that does not meet certain preset conditions is marked. This marking action alerts users to potentially problematic test item information, prompting further inspection and processing. By marking test item information that does not meet preset conditions, users can verify and correct it, thereby improving the accuracy of the test item information.

[0182] In practical applications, one or more post-processing operations such as deduplication, merging, and tagging can be combined to improve the quality and usability of test project information, depending on actual needs.

[0183] In some embodiments, the deduplication operation on the test item information of the test item includes at least one of the following:

[0184] Deduplication is performed based on the vector cosine similarity between multiple test items of the same test item;

[0185] A comparison model is used to deduplicate test item information for the same test item.

[0186] Deduplication can be performed based on the cosine similarity of vectors between test item information. Items with a similarity exceeding a similarity threshold are considered similar. Specifically, the test item information can first be converted into vector representations. Word embedding techniques from natural language processing (such as Word2Vec, GloVe, BERT, etc.) can be used to convert the text content of the test item information into vectors. Then, the cosine similarity between different test item information vectors is calculated. The higher the cosine similarity value, the more similar the two vectors are. A similarity threshold (e.g., 0.85) can be set; when the cosine similarity of the vectors of two test item information is greater than the threshold, the two test item information are considered duplicates, and only one of them can be retained.

[0187] Comparison models can be rule-based or machine learning-based, such as large language models. Rule-based models define rules to determine if two test item messages are duplicated, such as keyword matching or structural similarity. Machine learning-based models learn rules for determining duplicate test item messages by studying large amounts of sample data. Comparison models understand test item messages and identify differences between them. Specialized comparison models can be used to determine if two test item messages are duplicated.

[0188] For example, input the information of two candidate test items (including the name, description, and test parameters of the test items) into the large language model, and directly ask, "Do these two test item information describe the same technical concept or test requirement?" The large language model's understanding ability can usually make a more accurate judgment.

[0189] One or more deduplication methods can be combined to achieve the best deduplication results, depending on the actual needs. For example, vector cosine similarity can be used for initial deduplication, followed by a comparison model for fine-tuning. Deduplicating test item information can prevent duplicate test item information from interfering with subsequent processing and improve the quality of test item information.

[0190] When merging multiple test item information, you can select one of the multiple test item information as the base test item information for merging, and then merge the other test item information into the base test item information. The specific merging method is as follows.

[0191] In some embodiments, the merging of test project information for the same test project includes:

[0192] Retain the first test item information with the highest priority among multiple test item information for the same test item;

[0193] The second test item information from the plurality of test item information is merged into the first test item information to obtain the merged test item information;

[0194] The priority of each test item information among the plurality of test item information is determined based on at least one of the following: the type of reference document corresponding to each test item information; the amount of information contained in each test item information.

[0195] Specifically, the priority of each test item information can be determined first. Priority can be based on the type of reference document corresponding to the test item information; for example, internal documents take precedence over competitor documents, standard documents over product manuals, and product manuals over test reports. Priority can also be determined based on the amount of information contained in the test item information; for example, test item information containing complete information such as test parameters and judgment conditions has higher priority than test item information containing only partial information. After determining the priority, the test item information with the highest priority is retained as the first test item information.

[0196] Once the first test item information is determined, the second test item information can be merged into the first test item information. Specifically, information in the second test item information that is not duplicated in the first test item information can be identified and added to the first test item information. For example, if the first test item information contains test parameters but not decision conditions, while the second test item information contains decision conditions, then the decision conditions from the second test item information can be added to the first test item information.

[0197] In practical applications, different priorities for different types of reference documents or different types of test project information can be set according to actual needs to achieve the best merging effect. By merging multiple test project information for the same test project, more complete test project information can be obtained, improving the quality and usability of the test project information.

[0198] In some embodiments, the method further includes: establishing a tree structure containing multiple test items, wherein each node of the tree structure corresponds to at least one test item.

[0199] Here, the tree structure of the test items can be used to display them to users so that they can determine the hierarchy of the test items.

[0200] A tree structure is a hierarchical data structure consisting of nodes and edges connecting the nodes. Each node in a tree structure corresponds to a test item, and the connections between nodes represent the hierarchical relationships between test items. For example, a parent node can represent a test category, and its child nodes can represent the specific test items within that category.

[0201] In one possible implementation, the name of each node could be represented by the name of the test project.

[0202] Specifically, the hierarchical relationship between test items can be determined first. This can be based on the hierarchical relationship of test items in the reference document, or on the functional relationships or other relationships between test items. Then, based on the hierarchical relationship between test items, organize the test items into a tree structure.

[0203] A tree structure can be represented in multiple ways. For example, it can be represented using nested data structures (such as JSON objects) or graphical representations (such as tree diagrams).

[0204] In one possible implementation, the tree structure can be applied to measurement devices such as oscilloscopes, which can then perform test items based on the tree structure.

[0205] In this way, by establishing a tree structure, the relationships between test items can be displayed more intuitively, making it easier to understand and manage them. For example, test items can be categorized, filtered, queried, and modified based on the tree structure, improving the efficiency of test item management.

[0206] The specific method for establishing a tree structure is as follows.

[0207] In some embodiments, establishing a tree structure containing multiple test items includes at least one of the following: determining the level of the test item in the tree structure based on the hierarchical relationship of the test item in the corresponding reference document; determining the level of the test item in the tree structure based on a judgment model; determining the level of the test item in the tree structure based on predetermined rules; and setting at least one root node based on the standard associated with the target signal and / or the function associated with the target signal.

[0208] When creating a tree structure, you can first create an empty tree structure, then determine the hierarchy of different test items, and add the different test items to the tree structure.

[0209] Once the empty tree structure is determined, the standards and / or functions associated with the target signal can be used as the root nodes of the tree structure. Here, all test items for the target signal can be built into separate tree structures based on the standards and functions associated with the target signal. For example, if the target signal is a bus signal, the bus standard (such as USB, PCIe, etc.) can be used as the root node; if the target signal includes multiple signals (data signals, clock signals), the data signals and clock signals can be used as root nodes respectively.

[0210] To determine the hierarchy of test items, the structure of the reference document can be analyzed to identify the hierarchical relationship of the test items within the document. For example, if the reference document is a standard document, the hierarchy of the test items in the tree structure can be determined based on their chapter hierarchy within the standard document. If test item A is a sub-chapter of test item B in the reference document, then in the tree structure, test item A can be considered a child node of test item B.

[0211] In one possible implementation, the structure of the reference document may be determined during the parsing of the reference document and / or the recognition of the language using a language recognition model.

[0212] For example, if category information or hierarchy hints (such as a category field, or names containing separators such as ":", "-") are obtained during reference document parsing or large language model extraction, this information is used first to construct the initial parent-child relationship.

[0213] To determine the hierarchy of test items, a judgment model can be used to identify their position within a tree structure. This model can employ specialized judgment models to define the hierarchical relationships between test items. These models may utilize machine learning models, such as large language models. By learning from a large amount of sample data, the judgment model can automatically learn rules for determining the hierarchical relationships between test items.

[0214] For example, for test items with unclear relationships, a large language model can be used. For instance, provide the large language model with a list of test item names and instruct it to organize the test items into a logical hierarchical structure that reflects their subordinate relationships. The large language model then determines the hierarchical relationships.

[0215] The hierarchy of test items can be determined based on predefined rules. For example, certain predefined rules can be set to determine the hierarchy of test items in the tree structure. A shorter, more general name might be the parent node in the tree structure, while a node containing the parent node's name might be a child node.

[0216] In this way, by establishing a tree structure, the relationships between test items can be displayed more intuitively, making it easier to understand and manage them. For example, test items can be categorized, filtered, queried, and modified based on the tree structure, improving the efficiency of test item management.

[0217] In some embodiments, the method further includes: setting a corresponding identifier for each node of the tree structure, and using the connection relationship information of the identifier to characterize the tree structure.

[0218] In setting an identifier for each node in the tree structure, a unique identifier can be assigned to each test item. When representing the tree structure using the connection information of the identifiers, the hierarchical relationships between test items can be represented by the connections between the identifiers.

[0219] In one possible implementation, the identifier of the parent node can be set for each node. In this way, a complete tree structure can be formed through the hierarchical relationship of the identifiers.

[0220] In one possible implementation, the connection relationship information can be represented as a tree structure using an adjacency list structure or a nested object structure. An adjacency list can represent a tree structure by recording the direct child nodes of each node. A nested object structure directly represents the hierarchical relationship of the tree through object nesting.

[0221] In one possible implementation, connection information can characterize the depth of a node in the tree structure.

[0222] By using identifiers to represent nodes and connection information to represent the tree structure, complex tree structures can be simplified into a combination of identifiers and connections, facilitating storage, transmission, and processing. It also facilitates operations on the tree structure, such as adding, deleting, and searching for nodes.

[0223] In one possible implementation, a tree structure of the test items can be displayed in the interactive interface.

[0224] Specifically, the interactive interface clearly displays hierarchical relationships and supports expanding or collapsing the tree structure. The names of the test items are displayed on the tree structure nodes. When a node is selected, the corresponding details panel in the tree structure displays complete test item information (description, test parameters, judgment conditions, and source information).

[0225] In one possible implementation, the source information could be designed as a clickable link or marker that displays the source document name and location when hovered over or clicked (and even attempts to jump to the corresponding location in the preview document).

[0226] In one possible implementation, the user interface could provide the ability to edit the nodes in the tree structure.

[0227] For example, the interactive interface can provide inline editing (double-clicking the node name) or a pop-up editing box. It allows modification of the node's test item name, description, test parameters, and / or judgment conditions. Test parameters can be implemented as an add-delete-modify list. It provides "Add Child Item" and "Delete Item" buttons or a right-click menu. It supports dragging and dropping nodes to adjust their parent-child relationships or sibling order.

[0228] After editing the tree structure, users can save the modified tree structure data by clicking the "Save Edits" button.

[0229] In one possible implementation, the tree structure can be exported to a file in a predefined format. This predefined format could include CSV, etc. The file in the predefined format can then be used for downloading by users.

[0230] Specifically, depth-first search (DFS) or breadth-first search (BFS) is typically used to traverse the tree structure. Generate CSV rows: For each node, generate a CSV row containing at least one of the following: the node's unique identifier, the parent node's identifier, the node's depth in the tree, the test item name, description, test parameters, decision criteria, and source information.

[0231] The following provides several specific examples in conjunction with any of the above embodiments:

[0232] This example provides an interactive system for determining a test item's method. The interactive system can be implemented using a combination of hardware and software. A specific interactive system is as follows: Figure 2 As shown, it includes: the presentation layer, the application layer, and the data and service layer.

[0233] The presentation layer uses the React.js framework in conjunction with the Ant Design component library to build the front-end application: the presentation layer is responsible for user interaction, data visualization, and sending requests to the back-end.

[0234] The core components of the presentation layer include:

[0235] 1. App.js: Overall layout and routing.

[0236] 2. DocumentManager.js: Handles document upload, list display, and deletion logic, and calls the upload / document list API. It uses React-Dropzone and the Antd Upload / List component.

[0237] 3. AnalysisController.js: Contains a "Start Analysis" button, collects the IDs of the documents to be analyzed, and calls the analysis API.

[0238] 4. ResultViewer.js: The core display area. It uses the Antd Tree component to receive and render tree data. It handles node expand / collapse events.

[0239] 5. NodeEditor.js: When a user edits a node, it might display a pop-up Modal or a side Panel containing a form for modifying node attributes. It uses Antd Form, Input, TextArea, etc.

[0240] 6. ActionButtons.js: Contains buttons such as "Export CSV" and "Save Edits", which call the corresponding APIs.

[0241] 7. State Management: Use the React Context API or Redux / Zustand to manage global state, such as document lists, analysis result tree data, and editing state.

[0242] App.js, DocumentManager.js, AnalysisController.js, ResultViewer.js, NodeEditor.js, and ActionButtons.js are the front-end components.

[0243] The application layer uses Node.js and Express.js to build the backend application. The application layer is responsible for handling business logic, data processing, and interaction with the LLM and database.

[0244] The core components of the application layer include:

[0245] 1. API Endpoints (Express Routes): Define RESTful API interfaces for front-end calls.

[0246] 2. Controllers: Handle API requests and call the Service layer.

[0247] 3. Services: Implement core business logic.

[0248] DocumentService: Handles file storage, metadata management, and calls to parsing libraries.

[0249] AnalysisService: Coordinates the analysis process -> Calls DocumentService to obtain text -> Calls LLMService to extract -> Calls IntegrationService to integrate.

[0250] LLMService: Encapsulates the interaction with the LLM API, including prompt building, API calls, result parsing, and error handling. It requires configuration with axios or node-fetch to send HTTP requests.

[0251] IntegrationService: Implements deduplication, merging, and hierarchical construction algorithms.

[0252] ExportService: Implements the logic for traversing the tree and generating a CSV file.

[0253] Middleware: May include middleware such as authentication and authorization, request logging, and error handling.

[0254] API Endpoints (Express Routes) and Controllers are the controllers. DocumentService, AnalysisService, LLMService, IntegrationService, ExportService, and Middleware are the service layer components.

[0255] Data and Service Layer

[0256] The core components of the data and service layer include:

[0257] 1. File storage. Storage locations can include local file systems or cloud storage.

[0258] 2. Database (optional but recommended):

[0259] MongoDB (NoSQL): Suitable for storing JSON documents with flexible structures (such as analysis result trees and document metadata).

[0260] PostgreSQL (SQL): It can also be used to store tree structures or design relational tables via JSONB fields.

[0261] Stored content includes: user information, document metadata, analysis task status, and a persistent analysis results tree.

[0262] 3. External LLM Services: Such as OpenAI API (GPT-4, GPT-3.5-turbo), AnthropicClaude, Google Gemini, etc. Access requires authentication via their provided API Key.

[0263] 4. Document parsing libraries (integration): pdf.js, mammoth.js, xlsx.js, csv-parser, etc. are integrated as dependency libraries in the backend service.

[0264] Figure 2 In the diagram, dashed arrows represent API requests, while solid arrows represent internal calls.

[0265] In conjunction with the interactive system described above, this example also provides a method for determining test items, such as... Figure 3 As shown, the methods for determining test items include:

[0266] Step 301: User Interaction and Document Upload

[0267] 1. Users access the system through a web interface.

[0268] 2. The interface provides clearly defined areas for uploading "Internal Consistency Analysis Documents" (such as standards and internal specifications) and "Competitor Documents." In other words, it's for uploading reference documents.

[0269] 3. Users select local reference documents (supporting common formats such as PDF, DOCX, XLSX, and CSV) and upload them via drag-and-drop or a file selection dialog box.

[0270] 4. The front-end performs basic file type and size validation. File uploads are sent to the back-end via API POST / api / upload / consistency-doc or POST / api / upload / competitor-doc.

[0271] 5. The backend receives the file, processes it using libraries such as Multer, stores the file in a specified location on the server (or cloud storage), and records the file information (ID, name, type, source tag, upload time, storage path) in the database (or metadata file).

[0272] 6. The front-end retrieves the list of uploaded documents via the GET / api / documents API and displays it in the "Document Management Area," providing preview (if implemented, such as PDF preview) and delete (DELETE / api / documents / :id) functions.

[0273] Step 302: Trigger the analysis task:

[0274] 1. Users select one or more documents from the document list for analysis.

[0275] 2. Click the "Start Analysis" button. The front end sends a POST / api / analyze request to the back end. The request body contains a list of IDs of the documents to be analyzed.

[0276] Step 303: Backend - Document Parsing and Preprocessing:

[0277] 1. After receiving the analysis request, the backend finds the corresponding reference file based on the document ID.

[0278] 2. Format parsing: Call the corresponding parsing library for each supported format:

[0279] PDF: Using pdf.js. This solution needs to handle both text-based PDFs and image-based PDFs (the latter requires OCR; however, this solution can initially assume text-based PDFs, or use OCR as an optional preprocessing step). pdf.js can extract text content, font information, and certain layout / structure information (such as titles and lists).

[0280] Word (.docx): Use mammoth.js. It can convert DOCX to HTML or plain text, and can better preserve the structure of headings, lists, tables, etc.

[0281] Excel (.xlsx): Uses xlsx.js. It can read the contents of workbooks, worksheets, and cells, convert them into JSON arrays or objects, and preserve the row and column structure.

[0282] CSV: Use csv-parser. Parse a CSV file into an array of JSON objects.

[0283] 3. Content Extraction and Standardization: The parsed content is uniformly converted into a plain text sequence, while retaining or extracting structural information (such as headings, levels, and table labels) as much as possible. Irrelevant headers, footers, formatting characters, and other noise are removed.

[0284] 4. Text Chunking: Because LLM inputs typically have length limitations (Context Window), long documents need to be segmented into meaningful text chunks. Strategies can include:

[0285] Segmented by chapter / title (using structural information obtained during parsing).

[0286] Divide into paragraphs.

[0287] Fixed-size blocks (e.g., 1000-2000 tokens) may cut off semantics.

[0288] Optimal semantic segmentation: Segmentation is performed by combining structural information and semantic coherence, ensuring that each block contains relatively complete context. Each block needs to record its start and end positions in the original text (page number, paragraph number, etc.).

[0289] Step 304: Extracting test items from the backend-LLM driver:

[0290] 1. Iterative processing of text blocks: For each text block (i.e., sub-block) in each reference document, perform the following operations:

[0291] 2. Dynamic Prompt Construction: Based on a predefined Prompt template, a specific Prompt is generated by combining it with the content of the current text block.

[0292] LLM API call: Send the constructed Prompt to the configured LLM service (by calling the OpenAI API or similar interface via HTTPS, with the API Key included).

[0293] Result parsing and validation: Receive the JSON response returned by the LLM. Parse the JSON and validate its structure to ensure it meets expectations.

[0294] Handle potential API errors, timeouts, or invalid responses (such as LLM returning non-JSON text).

[0295] Source information association: Each successfully extracted test item is strictly associated with its source information (document ID, text block index, original location information, source_reference provided by LLM).

[0296] Step 305: Backend - Multi-source information integration, deduplication, and conflict handling:

[0297] 1. Collect all extraction results: Summarize the test items (and their sources) from all documents and all text blocks into a single list.

[0298] 2. Deduplication: This is a crucial step and requires a robust strategy.

[0299] Candidate test item information pair generation: Initial screening can be performed based on the test item name (such as using edit distance or Jaccard similarity) to find candidate test name information pairs that may have duplicates.

[0300] Precise similarity determination: Performing a more refined comparison of candidate test item information pairs.

[0301] Option A: Embedded Vector Similarity: Use a pre-trained sentence embedding model (such as Sentence-BERT) to convert the names and descriptions in the test item information into vectors, and calculate the cosine similarity. Set a threshold (such as 0.85) to determine whether they are similar.

[0302] Option B: LLM-assisted judgment: Feed the LLM with the names, descriptions, parameters, and other information of the two candidate test items, and directly ask, "Do these two test items describe the same technical concept or test requirement?" The LLM's comprehension ability usually allows for a more accurate judgment.

[0303] Hybrid approach: Use a combination of methods, such as using vector similarity for quick filtering and then using LLM to confirm ambiguous cases.

[0304] 3. Merging: For test items determined to be duplicates:

[0305] Select the main entry (first test item information): You can select the entry based on a certain priority (e.g., internal documents take precedence over competitor documents) or retain the entry with the most complete information.

[0306] Merge information:

[0307] Name: Choose the most standard or commonly used name.

[0308] Description: Combine or select a more detailed description.

[0309] Parameters / Limits: If parameters / limits differ from different sources, they need to be recorded, possibly displayed side-by-side, with the source indicated (e.g., Param: Voltage; Value: 800mV (Source: DocA), 750mV (Source: ...).

[0310] This is especially important for competitive analysis.

[0311] Merge Sources: Aggregate all indexes from the original sources into the merged entries.

[0312] 4. Conflict handling: Conflicts that cannot be automatically merged (such as contradictory parameter definitions) can be marked and left for the user to handle during the editing stage.

[0313] Step 306: Backend - Hierarchical Structure Construction:

[0314] 1. Initialization: Create an empty tree structure.

[0315] 2. Utilize explicit structures: If category information or hierarchy hints (such as a category field, or names containing separators such as "::", "-") are obtained during document parsing or LLM extraction, prioritize using this information to build initial parent-child relationships.

[0316] 3. Using LLM for inference: For test items with unclear relationships, LLM can be used again. For example, provide a set of test item names to the LLM and ask: "Please organize the following test items into a reasonable hierarchical structure that reflects their subordinate relationships."

[0317] 4. Heuristic rules: Apply some rules, such as shorter, more general names are likely to be parent nodes, and nodes containing the parent node name are likely to be child nodes.

[0318] 5. Root node processing: It may be necessary to define one or more top-level root nodes (e.g., by standard or by functional domain).

[0319] 6. ID Assignment: Assign a unique ID to each node and set its Parent_ID to form a complete tree representation (adjacency list or nested object structure). Calculate the Level (depth) of each node.

[0320] Step 307: Front-end and back-end interaction - result display and user editing:

[0321] 1. The backend returns the constructed tree-structured data (usually in JSON format, such as nested objects or adjacency lists) to the frontend via API GET / api / analysis-result / :id (assuming each analysis task has an ID).

[0322] 2. The front end receives the data and uses the **Ant Design Tree** component to render it into an interactive tree view.

[0323] 3. Presentation:

[0324] Clearly displays hierarchical relationships and supports expanding / collapse.

[0325] The node displays the name of the test item.

[0326] When a node is selected, its full information (description, parameter list, limits, and aggregated source index) is displayed in the adjacent details panel.

[0327] The source index can be designed as a clickable link or marker, displaying the source document name and location when hovered over or clicked.

[0328] (It can even attempt to jump to the corresponding location in the preview document, which requires front-end PDF library support).

[0329] 4. Editing:

[0330] Provides inline editing (double-click the node name) or a pop-up editing box.

[0331] Allows modification of a node's Name, Description, Parameters, and Limits. Parameters can be implemented as a list that can be added, deleted, and modified.

[0332] Provide buttons or right-click menus for "Add Sub-item" and "Delete Sub-item".

[0333] (Advanced) Supports dragging and dropping nodes to adjust their parent-child relationships or sibling order.

[0334] 5. Save: After editing, the user clicks the "Save Edits" button. The frontend sends the modified tree structure data (or only the changed parts) to the backend via PUT / api / analysis-result / :id. The backend updates the data in storage.

[0335] Step 308: Backend - Result Export:

[0336] 1. When a user clicks the "Export CSV" button, the front end triggers a POST / api / export request (which may include the analysis result ID).

[0337] 2. The backend retrieves the latest (potentially edited) tree structure data.

[0338] 3. Traversing tree structures: Trees are typically traversed using depth-first search (DFS) or breadth-first search (BFS).

[0339] 4. Generate CSV rows: For each node, generate one CSV row containing the following:

[0340] ID: A unique identifier for a node.

[0341] Parent_ID: The ID of the parent node (empty or 0 for the root node).

[0342] Level: The depth of a node in the tree.

[0343] Name: Test item name.

[0344] Description: Test item description.

[0345] Parameters: Formats the parameter array into a readable string (e.g., param1 = value1; param2 = value2).

[0346] Limits: The formatted limit string.

[0347] Source: Formats the list of source indices as a string (e.g., DocA:p5; CompB:Sec3.2).

[0348] 5. Return File: Return the generated CSV content as a file stream to the front end, triggering the browser to download it.

[0349] This application provides a test item determination device, such as... Figure 4 As shown, the test item determination device 40 includes a processing module 41, wherein the processing module is used for:

[0350] Obtain at least one reference document associated with the signal testing of the target signal;

[0351] A language recognition model is used to identify the at least one reference document, and test item information of the target signal associated with the test item is extracted from the at least one reference document.

[0352] In some embodiments, the processing module is further configured to: preprocess the at least one reference document;

[0353] The processing module is specifically used to: identify the at least one reference document that has undergone preprocessing using the language recognition model;

[0354] The processing module is specifically used for at least one of the following:

[0355] The reference document is parsed using a parsing library corresponding to its format to obtain a reference document in a predetermined format;

[0356] The reference document is converted into a text sequence of reference documents;

[0357] The reference document is segmented using a predetermined segmentation method, wherein the sub-blocks of the reference document obtained by the segmentation are respectively recognized by the language recognition model.

[0358] In some embodiments, the processing module is specifically used for at least one of the following:

[0359] Divide according to the chapters of the referenced document;

[0360] Divide according to the title of the referenced document;

[0361] Divide according to the paragraphs of the reference document;

[0362] The reference document is segmented according to a predetermined data volume;

[0363] The reference document is segmented according to its semantics.

[0364] In some embodiments, the processing module is further configured to perform at least one of the following:

[0365] Perform a deduplication operation on the test item information of the test items;

[0366] Merge test project information for the same test project;

[0367] Mark test items that do not meet the preset conditions.

[0368] In some embodiments, the processing module is specifically used for at least one of the following:

[0369] Deduplication is performed based on the vector cosine similarity between multiple test items of the same test item;

[0370] A comparison model is used to deduplicate test item information for the same test item.

[0371] In some embodiments, the processing module is specifically used for:

[0372] Retain the first test item information with the highest priority among multiple test item information for the same test item;

[0373] The second test item information from the plurality of test item information is merged into the first test item information to obtain the merged test item information;

[0374] The priority of each test item information among the plurality of test item information is determined based on at least one of the following: the type of reference document corresponding to each test item information; the amount of information contained in each test item information.

[0375] In some embodiments, the processing module is further configured to: establish a tree structure containing multiple test items, wherein each node of the tree structure corresponds to at least one test item.

[0376] In some embodiments, the processing module is specifically used for at least one of the following:

[0377] Based on the hierarchical relationship of the test items in the corresponding reference documents, determine the level of the test items in the tree structure;

[0378] The level of the test item in the tree structure is determined based on the judgment model;

[0379] The hierarchy of the test items in the tree structure is determined based on predetermined rules;

[0380] At least one root node is set based on the standard and / or the function of the target signal association.

[0381] In some embodiments, the processing module is further configured to:

[0382] Each node of the tree structure is assigned a corresponding identifier, and the tree structure is characterized by the connection relationship information of the identifiers.

[0383] In some embodiments, the test item information is used to indicate at least one of the following:

[0384] The name of the test item;

[0385] The settings for the test items;

[0386] The test parameters of the test items;

[0387] The judgment criteria for the test items;

[0388] The source information of the test items.

[0389] It should be understood that the division of units or modules in the above device is only a logical functional division. In actual implementation, they can be fully or partially integrated into a single physical entity, or they can be physically separated. Furthermore, the units or modules in the device can be implemented by a processor calling software: for example, the device includes a processor connected to a memory containing instructions. The processor calls the instructions stored in the memory to implement any of the above methods or to implement the functions of the units or modules in the above device. The processor can be, for example, a general-purpose processor, such as a Central Processing Unit (CPU) or a microprocessor, and the memory can be internal or external to the device. Alternatively, the units or modules in the device can be implemented in the form of hardware circuits. The functionality of some or all of the units or modules can be achieved through the design of these hardware circuits, which can be understood as one or more processors. For example, in one implementation, the hardware circuit is an application-specific integrated circuit (ASIC). The functionality of some or all of the units or modules is achieved through the design of the logical relationships between the components within the circuit. In another implementation, the hardware circuit can be implemented using a programmable logic device (PLD). Taking a field-programmable gate array (FPGA) as an example, it can include a large number of logic gates. The connection relationships between the logic gates are configured through configuration files, thereby achieving the functionality of some or all of the units or modules. All units or modules of the above device can be implemented entirely through processor-called software, entirely through hardware circuits, or partially through processor-called software with the remaining parts implemented through hardware circuits.

[0390] In this disclosure, the processor is a circuit with signal processing capabilities. In one implementation, the processor can be a circuit with instruction read and execute capabilities, such as a Central Processing Unit (CPU), a microprocessor, a graphics processing unit (GPU) (which can be understood as a type of microprocessor), or a digital signal processor (DSP). In another implementation, the processor can implement certain functions through the logical relationships of hardware circuits. The logical relationships of the aforementioned hardware circuits are fixed or reconfigurable. For example, the processor is a hardware circuit implemented using an application-specific integrated circuit (ASIC) or a programmable logic device (PLD), such as an FPGA. In a reconfigurable hardware circuit, the process of the processor loading a configuration document and configuring the hardware circuit can be understood as the process of the processor loading instructions to implement the functions of some or all of the above units or modules. In addition, it can also be a hardware circuit designed for artificial intelligence, which can be understood as an ASIC, such as a Neural Network Processing Unit (NPU), a Tensor Processing Unit (TPU), a Deep Learning Processing Unit (DPU), etc.

[0391] Figure 5 This is a schematic diagram of the structure of the electronic device 9100 provided in this embodiment. The electronic device 9100 can be a measuring device such as an oscilloscope or a logic analysis instrument, a terminal (e.g., user equipment), a chip, chip system, or processor that supports the implementation of any of the above methods, or a chip, chip system, or processor that supports the implementation of any of the above test item determination methods in the terminal. The electronic device 9100 can be used to implement the test item determination methods described in the above method embodiments; for details, please refer to the descriptions in the above method embodiments.

[0392] like Figure 5 As shown, the electronic device 9100 includes one or more processors 9101. The processor 9101 can be a general-purpose processor or a special-purpose processor, such as an oscilloscope processor or a central processing unit (CPU). The oscilloscope processor can be used for data processing, while the CPU can be used for data control, program execution, and processing of program data. The processor 9101 is used to invoke instructions to cause the electronic device 9100 to execute any of the above-mentioned test item determination methods.

[0393] In some embodiments, the electronic device 9100 further includes one or more memories 9102 for storing instructions. Optionally, all or part of the memories 9102 may also be located outside the electronic device 9100.

[0394] In some embodiments, the electronic device 9100 further includes one or more transceivers 9103. When the electronic device 9100 includes one or more transceivers 9103, the steps of sending, receiving and / or acquiring in the above method are performed by the transceivers 9103, and the other steps are performed by the processor 9101.

[0395] In some embodiments, the acquisition steps in the above method can also be executed by the processor 9101, for example, acquiring information from the memory 9102.

[0396] In some embodiments, a transceiver may include a receiver and a transmitter, which may be separate or integrated. Optionally, the terms transceiver, transceiver unit, transceiver, transceiver circuit, etc., may be used interchangeably; the terms transmitter, transmitting unit, transmitter, transmitting circuit, etc., may be used interchangeably; and the terms receiver, receiving unit, receiver, receiving circuit, etc., may be used interchangeably.

[0397] Optionally, the electronic device 9100 further includes one or more interface circuits 9104 connected to the memory 9102. The interface circuits 9104 can be used to receive signals from the memory 9102 or other devices, and can be used to send signals to the memory 9102 or other devices. For example, the interface circuits 9104 can read instructions stored in the memory 9102 and send the instructions to the processor 9101.

[0398] The electronic device 9100 described in the above embodiments may be a network device or a terminal, but the scope of the electronic device 9100 described in this disclosure is not limited thereto, and the structure of the electronic device 9100 may vary. Figure 5 The limitations. Electronic devices can be standalone devices or part of a larger device. For example, the electronic devices can be: (1) standalone integrated circuits (ICs), or chips, or chip systems or subsystems; (2) a collection of one or more ICs, optionally including storage components for storing data or programs; (3) ASICs, such as modems; (4) modules that can be embedded in other devices; (5) receivers, terminal devices, smart terminal devices, cellular phones, wireless devices, handheld devices, mobile units, vehicle-mounted devices, network devices, cloud devices, artificial intelligence devices, etc.; (6) others, etc.

[0399] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program commands. The aforementioned program can be stored in a storage medium, including various media capable of storing program code such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0400] Alternatively, if the integrated units described above are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several commands to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

[0401] It should be understood that the above embodiments are exemplary and are not intended to encompass all possible implementations included in the claims. Various modifications and changes can be made to the above embodiments without departing from the scope of this disclosure. Similarly, the various technical features of the above embodiments can be arbitrarily combined to form other embodiments of the present invention that may not be explicitly described. Therefore, the above embodiments only illustrate several implementations of the present invention and do not limit the scope of protection of this patent.

Claims

1. A method for determining test items, characterized in that, The method includes: Obtain at least one reference document associated with the signal testing of the target signal; A language recognition model is used to identify the at least one reference document, and test item information of the target signal associated with the test item is extracted from the at least one reference document.

2. The method according to claim 1, characterized in that, The method further includes: preprocessing the at least one reference document; The step of using a language recognition model to identify the at least one reference document includes: using the language recognition model to identify the at least one reference document that has undergone preprocessing; The preprocessing of the at least one reference document includes at least one of the following: The reference document is parsed using a parsing library corresponding to its format to obtain a reference document in a predetermined format; The reference document is converted into a text sequence of reference documents; The reference document is segmented using a predetermined segmentation method, wherein the sub-blocks of the reference document obtained by the segmentation are respectively recognized by the language recognition model.

3. The method according to claim 2, characterized in that, The segmentation of the reference document using a predetermined segmentation method includes at least one of the following: Divide according to the chapters of the referenced document; Divide according to the title of the referenced document; Divide according to the paragraphs of the reference document; The reference document is segmented according to a predetermined data volume; The reference document is segmented according to its semantics.

4. The method according to claim 1, characterized in that, The method further includes at least one of the following: Perform a deduplication operation on the test item information of the test items; Merge test project information for the same test project; Mark test items that do not meet the preset conditions.

5. The method according to claim 4, characterized in that, The deduplication operation on the test item information of the test items includes at least one of the following: Deduplication is performed based on the vector cosine similarity between multiple test items of the same test item; A comparison model is used to deduplicate test item information for the same test item.

6. The method according to claim 4, characterized in that, The test project information for merging the same test project includes: Retain the first test item information with the highest priority among multiple test item information for the same test item; The second test item information from the plurality of test item information is merged into the first test item information to obtain the merged test item information; The priority of each test item information among the plurality of test item information is determined based on at least one of the following: the type of reference document corresponding to each test item information; the amount of information contained in each test item information.

7. The method according to any one of claims 1 to 6, characterized in that, The method further includes: establishing a tree structure containing multiple test items, wherein each node of the tree structure corresponds to at least one test item.

8. The method according to claim 7, characterized in that, The establishment of a tree structure containing multiple test items includes at least one of the following: Based on the hierarchical relationship of the test items in the corresponding reference documents, determine the level of the test items in the tree structure; The level of the test item in the tree structure is determined based on the judgment model; The hierarchy of the test items in the tree structure is determined based on predetermined rules; At least one root node is set based on the standard and / or the function of the target signal association.

9. The method according to claim 7, characterized in that, The method further includes: Each node of the tree structure is assigned a corresponding identifier, and the tree structure is characterized by the connection relationship information of the identifiers.

10. The method according to any one of claims 1 to 6, characterized in that, The test item information is used to indicate at least one of the following: The name of the test item; The settings for the test items; The test parameters of the test items; The judgment criteria for the test items; Source information of the test items; The description information of the test items.

11. A test item determination device, characterized in that, The device includes a processing module, wherein the processing module is used for: Obtain at least one reference document associated with the signal testing of the target signal; A language recognition model is used to identify the at least one reference document, and test item information of the target signal associated with the test item is extracted from the at least one reference document.

12. An electronic device comprising a processor, a memory, and an executable program stored in the memory and executable by the processor, characterized in that, When the processor runs the executable program, it performs the steps of the test item determination method as described in any one of claims 1 to 10.

13. A storage medium having an executable program stored thereon, characterized in that, When the executable program is executed by a processor, it implements the steps of the test item determination method as described in any one of claims 1 to 10.

Images