Domestic artificial intelligence computing power adaptation test service platform

Abstract

The invention provides a domestic artificial intelligence computing power adaptation test service platform, which belongs to the technical field of data management, and comprises a user login module used for carrying out identity authentication and entering a platform main interface after the authentication is passed; a plurality of function options for a user to select are provided in the main interface, and each function option is used for executing preset logic when being selected so as to obtain an execution result and display the execution result; the question and answer module is used for receiving questions related to chips or manufacturers and proposed by users, converting the questions into question embedding vectors, generating cue words based on the question embedding vectors and embedding vectors matched with the question embedding vectors and retrieved from a preset data knowledge base, and sending the cue words to a server; and inputting the cue word and the question embedding vector into the large language model so as to feed back and obtain a question answer. The system has the beneficial effects that through integrated information management, comprehensive and accurate domestic chip information service is provided for enterprises, and the information retrieval efficiency is improved; and the use experience is improved through interactive questions and answers.

Description

Technical Field

[0001] The present invention relates to the field of data management technology, and in particular to a domestic artificial intelligence computing power adaptation testing service platform. Background Art

[0002] Comparative analysis of chip performance is a key step in evaluating the competitiveness of electronic products and selecting appropriate hardware solutions in the semiconductor industry. Currently, a variety of chip comparison software has emerged on the market, which helps users evaluate and compare the performance of different graphics processing units (GPUs) by providing detailed technical specifications, performance indicators and in-depth data analysis. Existing chip comparison software usually has a user-friendly interface, allowing users to select and compare various performance parameters of different GPUs through intuitive operations; at the same time, it can display key indicators including but not limited to core frequency, memory bandwidth, floating-point computing performance, etc.

[0003] However, the data contained in the existing chip comparison software lacks data support from domestic chip manufacturers, and the data range is limited; moreover, these software are limited to displaying the performance parameters of the chip, lack the generation and preview functions of test reports, and the functions are relatively simple; at the same time, the existing software platform has obvious deficiencies in user data security protection, cross-platform compatibility, and interactive question-and-answer queries, which can not effectively protect the security and privacy of user data, but also affect the user experience on different operating systems and devices, and lack convenience when executing queries. With the rapid development of the domestic chip industry, the market is in urgent need of a comprehensive software platform to centrally manage and comprehensively display detailed information on domestic chips and make up for the many defects of existing technologies. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides a domestic artificial intelligence computing power adaptation testing service platform.

[0005] The technical problem solved by the present invention can be achieved by adopting the following technical solutions:

[0006] A domestic artificial intelligence computing power adaptation test service platform, including:

[0007] The user login module is used to receive user information, perform identity authentication based on a preset whitelist, and enter the main interface of the platform after authentication;

[0008] The main interface provides a plurality of function options for the user to select, and each function option is used to execute a preset logic when selected to obtain and display an execution result;

[0009] The question-answering module is used to receive questions related to chips or manufacturers raised by users, convert the questions into question embedding vectors, and generate prompt words based on the question embedding vectors and embedding vectors matching the question embedding vectors retrieved from a preset data knowledge base, and input the prompt words and the question embedding vectors into a large language model to provide feedback to obtain answers to the questions.

[0010] Preferably, it also includes:

[0011] A first user monitoring and analysis module, used to monitor user behavior and status, and analyze the user behavior and status to obtain a first analysis result;

[0012] a whitelist updating module, connected to the first user monitoring and analysis module and the user login module respectively, and configured to adjust the entries of the whitelist according to the first analysis result;

[0013] The adjustment behaviors of the entries in the whitelist include automatic addition, automatic removal and temporary adjustment.

[0014] Preferably, the function options include: viewing option, query option, comparison option and preview option;

[0015] The viewing option is used to display the details page of the information card displayed on the service platform interface, and the information card includes performance parameters, graphics card model, enterprise information, and test report;

[0016] The comparison option is used to compare the selected comparison items according to the comparison dimensions and weights customized by the user, obtain the comparison results, and display the comparison results;

[0017] The query option is used to receive the query item entered by the user in the search keyword input box, and display the relevant information corresponding to the query item when the query item is one; and display the comparison results of the relevant information corresponding to two or more query items when the query items are two or more.

[0018] Preferably, it also includes:

[0019] A data collection module, which is connected to multiple data sources and is used to aggregate information from the data sources when the data sources publish new content to obtain aggregated data, wherein the aggregated data is marked with corresponding data sources;

[0020] A data classification module, connected to the data collection module, is used to classify the aggregated data according to preset categories to obtain classification results;

[0021] A vector conversion module, connected to the data classification module, used to convert the data source information of different categories in the classification result into an embedding vector using an embedding model to obtain an embedding vector of each category;

[0022] A data knowledge base updating module is connected to the data classification module and is used to update the embedding vector into the data knowledge base according to its classification.

[0023] Preferably, the data collection module comprises:

[0024] A multi-source verification unit, which is used to cross-check information from different data sources to filter out false, inaccurate or inconsistent information;

[0025] A data source marking unit, connected to the multi-source verification unit, for marking the data source information after filtering;

[0026] An aggregation unit is connected to the multi-source verification unit and is used to aggregate the filtered data source information and the corresponding data source annotations to obtain the aggregated data.

[0027] Preferably, the data source labeling unit is also used to compare the labeled data in the data knowledge base with the historical data source for verification and correction.

[0028] Preferably, the data knowledge base updating module includes:

[0029] A first updating unit, configured to add the embedding vector having a first preset attribute to the data knowledge base, while retaining the corresponding historical embedding vector;

[0030] The second updating unit is used to add or cover the embedding vector that does not have the first preset attribute to the data knowledge base.

[0031] Preferably, it also includes:

[0032] A second user monitoring and analysis module, used to monitor user preferences and usage habits, and analyze the user preferences and usage habits to obtain a second analysis result;

[0033] a display content adjustment module, connected to the second user monitoring and analysis module, and configured to adjust the display content according to the second analysis result;

[0034] The adjustment behavior of the displayed content includes display priority and arrangement mode.

[0035] Preferably, it also includes:

[0036] The field annotation display module is used to provide annotations for each data field in the service platform interface, and the annotations include multilingual annotations and semantic interpretations.

[0037] Preferably, the data knowledge base includes a performance parameter table, a graphics card model table, a company information table, and a test report table.

[0038] The advantages or beneficial effects of the technical solution of the present invention are:

[0039] The present invention provides functional options through integrated information management, and provides comprehensive and accurate domestic chip information services to enterprises on a unified platform, greatly improving the efficiency of information retrieval; through interactive questions and answers, users can easily browse, query and understand complex chip data, improving the overall user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a system architecture diagram of a domestic artificial intelligence computing power adaptation test service platform in a preferred embodiment of the present invention;

[0041] Figure 2 This is a system flow chart of a domestic artificial intelligence computing power adaptation test service platform in a preferred embodiment of the present invention;

[0042] Figure 3 FIG. 1 is a flow chart of a question-answering module in a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0043] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0045] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, but they are not intended to limit the present invention.

[0046] An embodiment of the present invention provides a domestic artificial intelligence computing power adaptation testing service platform, which aims to provide enterprises with comprehensive and accurate domestic chip information services.

[0047] The platform proposed in the embodiment of the present invention is a mobile phone APP developed based on PalmBao, which is specially designed for senior management of enterprises and government departments. It integrates advanced large language models and combines powerful knowledge base functions to provide comprehensive and accurate domestic chip information services.

[0048] The platform is an integrated information management platform that centrally manages and displays detailed information about domestic chips, covering company information, product parameters, test results, product comparisons and other internal information, and has become the core information release platform of the Adaptation Test Center. The platform uses the most advanced data processing and visualization technology to enable users to make scientific purchasing decisions and formulate technical strategic plans based on detailed data. It is not only an important decision-making auxiliary tool for corporate executives, but also provides accurate chip information for government commissions.

[0049] Figure 1 It is the system architecture diagram of the platform. The entire platform is divided into three parts: user interface 100, backend service 200, and database 300. User interface 100 includes main interface, chat information, graphics card information interface, and performance comparison interface. Backend service 200 includes query service, data management service, user authentication service, and comparison service. Database 300 is the data knowledge base below, including but not limited to performance parameter table, graphics card model table, enterprise information table, test report table, etc.

[0050] In a preferred embodiment of the present invention, based on the above problems existing in the prior art, a domestic artificial intelligence computing power adaptation test service platform is provided, including:

[0051] The user login module is used to receive user information, perform identity authentication based on a preset whitelist, and enter the main interface of the platform after authentication;

[0052] The main interface provides multiple function options for users to choose from. Each function option is used to execute the preset logic when selected to obtain and display the execution results;

[0053] The question-answering module is used to receive questions related to chips or manufacturers raised by users, convert the questions into question embedding vectors, and generate prompt words based on the question embedding vectors and embedding vectors matching the question embedding vectors retrieved from a preset data knowledge base, and input the prompt words and question embedding vectors into a large language model to provide feedback to obtain answers to the questions.

[0054] Specifically, in this embodiment, the user login module implements user identity authentication based on the above user authentication service. By adopting a whitelist registration mechanism, the accuracy and high level of the user group are ensured, focusing on providing reliable data support for decision makers.

[0055] Functional options include the above-mentioned query service, data management service, and comparison service.

[0056] The question-and-answer module uses the chat interface to connect to the large language model interface to answer chip-related questions. Based on the large language model, the manufacturer's chip data knowledge base is integrated to answer questions about specific manufacturers and chips. This module not only relies on the natural language processing capabilities of the large language model, but also combines a data knowledge base customized specifically for domestic chips. This knowledge base not only includes public technical information, but also integrates internal data and expertise in specific fields. The large language model extracts relevant information from the knowledge base through multi-level semantic understanding and context association to generate highly accurate answers.

[0057] Furthermore, the platform of the embodiment of the present invention pays special attention to deep information processing in the field, and uses customized semantic indexing and embedding technology, so that the model can not only understand general questions, but also handle professional questions involving complex technical details. Compared with the existing general question-answering system, the accuracy of answers to questions about specific manufacturers and chips is significantly improved.

[0058] As a preferred embodiment, the invention further comprises:

[0059] A first user monitoring and analysis module, used to monitor user behavior and status, and analyze the user behavior and status to obtain a first analysis result;

[0060] The whitelist updating module is connected to the first user monitoring and analysis module and the user login module respectively, and is used to adjust the entries of the whitelist according to the first analysis result.

[0061] The adjustment behaviors of whitelist entries include automatic addition, automatic removal, and temporary adjustment.

[0062] Specifically, considering the lack of real-time monitoring and feedback functions for user activities in the prior art, the embodiments of the present invention can dynamically adjust whitelist entries according to the user's behavior and status to implement a real-time update mechanism. For example, expired or inactive users are automatically removed, and users who meet new conditions are automatically added.

[0063] Furthermore, the platform provides a user-friendly interactive interface, and provides search, comparison and analysis tools. Functional options include but are not limited to viewing options, query options, comparison options and preview options to optimize the user experience.

[0064] As a preferred implementation, the function options include a viewing option, and the viewing option is used to display a detail page of an information card displayed on the service platform interface, and the information card includes performance parameters, graphics card model, enterprise information, and test report.

[0065] As a preferred embodiment, the function options include comparison options;

[0066] The comparison option is used to compare the selected comparison items according to the user-defined comparison dimensions and weights, obtain the comparison results, and display the comparison results.

[0067] Specifically, traditional comparison tools are usually fixed on several standard dimensions, such as core frequency, video memory capacity, power consumption, etc., and users cannot customize the comparison dimensions and weights according to specific needs. The comparison option function of the embodiment of the present invention allows users to customize the dimensions and weights of performance evaluation and comparison according to actual needs and specific application scenarios (such as graphics rendering, scientific computing, etc.), providing users with customized GPU performance evaluation results, making the comparison results more targeted and practical, and meeting the specific needs of different users.

[0068] The platform also summarizes test reports from specific manufacturers and provides a result comparison function to help users quickly understand the performance comparison of different products.

[0069] Furthermore, existing comparison tools are usually limited to comparing 2-3 GPUs, and have poor support for simultaneous comparison of multiple GPUs. In addition, the visualization is limited and usually presented in a table format, which is difficult for users to understand intuitively. The embodiments of the present invention support parallel comparison of up to 4 GPUs and provide multiple visualization display methods, such as radar charts, bar charts, line charts, etc., to help users intuitively see the performance of each GPU in different dimensions.

[0070] At the same time, the platform also supports users to freely switch visualization views, and allows users to dynamically add or remove comparison dimensions during the comparison process, improving the flexibility of comparison and user experience.

[0071] Users can also select and adjust different performance indicators, such as deep learning inference performance, game frame rate, energy efficiency, etc.

[0072] The embodiments of the present invention take into account the diverse demands of different users for GPU performance in different application environments, and can dynamically adjust the performance evaluation dimension based on real-time data.

[0073] As a preferred embodiment, the function options include query options;

[0074] The query option is used to receive the query item entered by the user in the search keyword input box, and to display the relevant information corresponding to the query item when there is one query item; and to display the comparison results of the relevant information corresponding to two or more query items when there are two or more query items.

[0075] As a preferred implementation, the function options include a preview option; the preview, zoom-in and zoom-out functions of PDF files are supported to enhance the user experience of viewing the test report.

[0076] As a preferred embodiment, the invention further comprises:

[0077] The data collection module is connected to multiple data sources and is used to aggregate information from the data sources when the data sources publish new content to obtain aggregated data, in which the corresponding data sources are marked;

[0078] A data classification module, connected to the data collection module, is used to classify the aggregated data according to preset categories to obtain classification results;

[0079] The vector conversion module is connected to the data classification module and is used to use the embedding model to convert the data source information of different categories in the classification results into embedded vectors to obtain the embedded vectors of each category for efficient storage and retrieval;

[0080] The data knowledge base update module is connected to the data classification module and is used to update the embedding vectors into the data knowledge base according to their classification, so as to continuously enrich and improve the data resources of the platform.

[0081] In this embodiment, the preset categories include but are not limited to different categories such as company information, product parameters, and test results.

[0082] In this embodiment, the data source includes but is not limited to the official website of the manufacturer, industry reports, patent databases, etc. By integrating official data manuals, an authoritative information source is provided.

[0083] Specifically, in this embodiment, natural language processing (NLP) and machine learning technologies are used to automatically aggregate information from multiple data sources, such as the manufacturer's official website, industry reports, patent databases, etc. Through intelligent classification algorithms, the collected data is automatically classified into different categories such as company information, product parameters, test results, etc., to ensure the real-time and accuracy of the information.

[0084] The embodiment of the present invention ensures that the information on the platform is real-time and accurate through an automated data update mechanism, reducing decision errors caused by outdated data. Different from traditional manual classification and static databases, this dynamic aggregation and classification technology can automatically adapt to changes in new data, significantly improving the efficiency and accuracy of data management.

[0085] As a preferred embodiment, the data collection module includes:

[0086] A multi-source verification unit, which is used to cross-check information from different data sources to filter out false, inaccurate or inconsistent information;

[0087] A data source marking unit, connected to the multi-source verification unit, is used to mark the data source information after filtering;

[0088] The aggregation unit is connected to the multi-source verification unit and is used to aggregate the filtered data source information and the corresponding data source annotations to obtain aggregated data.

[0089] Specifically, in this embodiment, the credibility of information is increased by clearly marking the source of data.

[0090] Furthermore, the platform can display internal information such as core personnel and major company events to increase information transparency. In order to further ensure the accuracy and reliability of the displayed internal information, the platform uses multi-source data verification to cross-check information from different sources to filter out possible false or inaccurate data. Compared with the traditional single data source display, the embodiment of the present invention innovatively combines the technology of multi-source data verification to further ensure the credibility of the information.

[0091] The data collection module also includes a multi-source verification unit and a data source annotation unit. The multi-source verification unit is responsible for cross-checking information from different data sources to filter out false, inaccurate or inconsistent information. The data source annotation unit annotates the filtered data source information to ensure that the source of the data is clear and traceable.

[0092] As a preferred implementation, the data source labeling unit is also used to compare the labeled data in the data knowledge base with the historical data source for verification and correction.

[0093] Specifically, in this embodiment, an automated verification and correction mechanism is introduced to automatically detect and correct possible annotation errors by comparing the annotated data with the historical data source information, thereby ensuring the accuracy and consistency of the annotation information.

[0094] As a preferred implementation, the data knowledge base update module includes:

[0095] A first updating unit, configured to add an embedding vector having a first preset attribute to a data knowledge base, while retaining a corresponding historical embedding vector;

[0096] The second updating unit is used to add or cover the embedding vector that does not have the first preset attribute to the data knowledge base.

[0097] In this embodiment, the first preset attribute includes but is not limited to the version of the information.

[0098] Specifically, in this embodiment, by introducing a version control system and an automatic update mechanism, it is ensured that the official data manual on the platform is kept synchronized with the latest official release.

[0099] When a new version of the data sheet is officially released, the platform will automatically download, parse and update the corresponding content, while retaining the historical version for user reference.

[0100] Compared with manual updates and single-version display, this automated version control and update mechanism improves the timeliness and reliability of data and reduces the risk of information lag.

[0101] As a preferred embodiment, the invention further comprises:

[0102] A second user monitoring and analysis module is used to monitor user preferences and usage habits, and analyze the user preferences and usage habits to obtain a second analysis result;

[0103] a display content adjustment module, connected to the second user monitoring and analysis module, and configured to adjust the display content according to the second analysis result;

[0104] The adjustment behavior of display content includes display priority and arrangement.

[0105] Specifically, the embodiment of the present invention adopts intelligent data display technology. Through data analysis and user behavior learning, the system can automatically adjust the priority and arrangement of displayed content according to user preferences and usage habits.

[0106] The embodiments of the present invention may also be combined with data visualization technology, and the system may also intuitively display complex data in various forms such as charts and graphs to enhance the user's understanding and decision-making capabilities.

[0107] As a preferred embodiment, the invention further comprises:

[0108] The field annotation display module is used to provide annotations for each data field in the service platform interface, including multilingual annotations and semantic interpretations.

[0109] Specifically, the embodiment of the present invention also supports field annotation and data field meaning display functions, which helps users understand the meaning of the data by providing annotations for each data field.

[0110] The platform also integrates multilingual support, which can automatically identify the user's language preference and provide annotations in the corresponding language. Through machine translation and pre-trained multilingual models, the system can provide accurate multilingual annotations for each field, improving the user experience.

[0111] Furthermore, the platform can also combine semantic network and knowledge graph technology to provide a deeper semantic interpretation for data fields. For example, for a "GPU power consumption" field, the system will not only annotate its basic meaning, but also provide background information and calculation methods related to power consumption, allowing users to have a deeper understanding of the actual meaning of the data field.

[0112] As a preferred implementation, the data knowledge base includes a performance parameter table, a graphics card model table, a company information table, and a test report table.

[0113] Figure 2 The entire system flow chart of the platform is shown. After the user logs in to the system, he enters the main interface. After that, the user can view the graphics card company information, browse the graphics card models or query performance parameters, compare graphics card performance, query test reports, and use the chat interface to ask questions. The user can also update data and set up the system.

[0114] Figure 3 The flowchart of the question-answering module is shown, which shows the complete process from data collection to question answering in the chip data question-answering process based on a large language model and knowledge base, involving multiple links such as data preprocessing, embedding vector generation, database storage, question processing, and answer generation. It is mainly divided into the following 12 steps:

[0115] S101. Obtain manufacturer and chip data:

[0116] Collect and organize data about chip manufacturers and the chips themselves.

[0117] S102, generating multiple sub-documents:

[0118] The collected data is divided into multiple sub-documents for subsequent processing and analysis.

[0119] S103. Call the Embedding model:

[0120] Use an embedding model to convert the text in the sub-document into an embedding vector, which is a technique for converting text information into a numerical vector and is commonly used in natural language processing.

[0121] S104. Generate text embedding vector:

[0122] Convert the text content of sub-documents into embedding vectors for further comparison and analysis.

[0123] S105. Store in PostGres:

[0124] The generated embedding vectors are stored in a PostgreSQL database for easy management and querying.

[0125] S106. Get relevant documents from PostGres:

[0126] Based on the user’s needs or questions, relevant embedding vectors are retrieved from the PostgreSQL database.

[0127] S107. User raises question:

[0128] Users ask questions related to chips or manufacturers through the interface.

[0129] S108. Call the Embedding model:

[0130] The embedding model is used again to convert the user’s question into an embedding vector.

[0131] S109. Generate the embedding vector of the problem:

[0132] Convert the user’s question text into an embedding vector so that it can be matched with the vectors in the database.

[0133] S110, generate prompt words:

[0134] Based on the user question and the retrieved embedding vector, relevant prompt words or keywords are generated to assist the large language model in providing more accurate answers.

[0135] S111. Calling the large language model:

[0136] Use a large language model to combine the generated prompt word and question embedding vectors for in-depth analysis and processing.

[0137] S112. Return the answer:

[0138] The system provides users with accurate and detailed answers based on the analysis results of the large language model.

[0139] Embodiment 1

[0140] Users register by receiving a verification code through their mobile phone number. On the registration interface, users need to fill in their user name and company information and submit it for review. After registration, the system submits the user information to the administrator for review. After the review is passed, the user can log in to the system. The user enters the mobile phone number and verification code to log in. After the system verifies successfully, the user enters the main interface. The homepage displays information cards of three popular companies, and users can click to enter the company details interface. The homepage displays information cards of four popular products, and users can click to view product details. The manufacturer details page displays the company's basic information, product list and test report. Display product series, users can click to view multiple GPUs in a product series, and expand the detailed data of the GPU. Users can select up to four GPUs for comparison, and the system displays the comparison results of key performance indicators such as computing power, video memory, and power consumption. The test report function can display test reports of specific manufacturers, classified by company and business. Click the test report to display detailed information such as test time, tester, and test status. Click the preview button to support the preview function of PDF files, and users can view the detailed content of the report.

[0141] Embodiment 2

[0142] Enter the My Dialogue interface, enter a specific manufacturer name or abbreviation, and the system will return detailed information about the company. Enter a specific chip product, and the system will return the specific parameters of the chip, including computing power, video memory, power consumption, chip information, form, card interconnection, codec, and comparison with NVIDIA products. Enter two product names and the system will return a comparison of the parameters of the two products.

[0143] The advantages or beneficial effects of adopting the above technical solution are:

[0144] (1) Through integrated information management, users can quickly obtain all domestic chip-related information on a unified platform, greatly improving the efficiency of information retrieval and acquisition;

[0145] (2) The automated data update mechanism ensures that the information on the platform is always up to date, reducing decision-making errors caused by outdated data and ensuring the real-time and accuracy of data;

[0146] (3) The user-friendly interactive interface and intuitive data analysis tools enable users to easily browse, query, and understand complex chip data, thereby improving the overall user experience;

[0147] (4) Advanced data analysis and report generation capabilities provide users with in-depth insights, helping them make more informed decisions and enhancing decision support capabilities.

[0148] (5) Advanced security technology protects user data and privacy, enhances user trust in the platform, and ensures data security and privacy.

[0149] The above are only preferred embodiments of the present invention, and are not intended to limit the implementation methods and protection scope of the present invention. Those skilled in the art should be aware that all solutions obtained by equivalent substitutions and obvious changes made using the contents of this specification and illustrations should be included in the protection scope of the present invention.

Claims

1. A domestic artificial intelligence computing power adaptation test service platform, characterized in that: include: The user login module is used to receive user information, perform identity authentication based on a preset whitelist, and enter the main interface of the platform after authentication; The main interface provides a plurality of function options for the user to select, and each function option is used to execute a preset logic when selected to obtain and display an execution result; The question-answering module is used to receive questions related to chips or manufacturers raised by users, convert the questions into question embedding vectors, and generate prompt words based on the question embedding vectors and embedding vectors matching the question embedding vectors retrieved from a preset data knowledge base, and input the prompt words and the question embedding vectors into a large language model to provide feedback to obtain answers to the questions.

2. The domestic artificial intelligence computing power adaptation test service platform according to claim 1 is characterized in that: Also includes: A first user monitoring and analysis module, used to monitor user behavior and status, and analyze the user behavior and status to obtain a first analysis result; a whitelist updating module, connected to the first user monitoring and analysis module and the user login module respectively, and configured to adjust the entries of the whitelist according to the first analysis result; The adjustment behaviors of the entries in the whitelist include automatic addition, automatic removal and temporary adjustment.

3. The domestic artificial intelligence computing power adaptation test service platform according to claim 1 is characterized in that: The function options include: viewing option, query option, comparison option and preview option; The viewing option is used to display the details page of the information card displayed on the service platform interface, and the information card includes performance parameters, graphics card model, enterprise information, and test report; The comparison option is used to compare the selected comparison items according to the comparison dimensions and weights customized by the user, obtain the comparison results, and display the comparison results; The query option is used to receive the query item entered by the user in the search keyword input box, and display the relevant information corresponding to the query item when the query item is one; and display the comparison results of the relevant information corresponding to two or more query items when the query items are two or more.

4. The domestic artificial intelligence computing power adaptation test service platform according to claim 1 is characterized in that: Also includes: A data collection module, which is connected to multiple data sources and is used to aggregate information from the data sources when the data sources publish new content to obtain aggregated data, wherein the aggregated data is marked with corresponding data sources; A data classification module, connected to the data collection module, is used to classify the aggregated data according to preset categories to obtain classification results; A vector conversion module, connected to the data classification module, used to convert the data source information of different categories in the classification result into an embedding vector using an embedding model to obtain an embedding vector of each category; A data knowledge base updating module is connected to the data classification module and is used to update the embedding vector into the data knowledge base according to its classification.

5. The domestic artificial intelligence computing power adaptation test service platform according to claim 4 is characterized in that: The data collection module includes: A multi-source verification unit, which is used to cross-check information from different data sources to filter out false, inaccurate or inconsistent information; A data source marking unit, connected to the multi-source verification unit, for marking the data source information after filtering; An aggregation unit is connected to the multi-source verification unit and is used to aggregate the filtered data source information and the corresponding data source annotations to obtain the aggregated data.

6. The domestic artificial intelligence computing power adaptation test service platform according to claim 5 is characterized in that: The data source labeling unit is also used to compare the labeled data in the data knowledge base with the historical data source for verification and correction.

7. The domestic artificial intelligence computing power adaptation test service platform according to claim 4 is characterized in that: The data knowledge base updating module comprises: A first updating unit, configured to add the embedding vector having a first preset attribute to the data knowledge base, while retaining the corresponding historical embedding vector; The second updating unit is used to add or cover the embedding vector that does not have the first preset attribute to the data knowledge base.

8. The domestic artificial intelligence computing power adaptation test service platform according to claim 1 is characterized in that: Also includes: A second user monitoring and analysis module, used to monitor user preferences and usage habits, and analyze the user preferences and usage habits to obtain a second analysis result; a display content adjustment module, connected to the second user monitoring and analysis module, and configured to adjust the display content according to the second analysis result; The adjustment behavior of the displayed content includes display priority and arrangement mode.

9. The domestic artificial intelligence computing power adaptation test service platform according to claim 1 is characterized in that: Also includes: The field annotation display module is used to provide annotations for each data field in the service platform interface, and the annotations include multilingual annotations and semantic interpretations.

10. The domestic artificial intelligence computing power adaptation test service platform according to claim 1 is characterized in that: The data knowledge base includes a performance parameter table, a graphics card model table, an enterprise information table, and a test report table.

Images