Bid evaluation method and system based on artificial intelligence

Abstract

The invention discloses a bid evaluation method and system based on artificial intelligence, and belongs to the technical field of bid procurement, and the method comprises the following steps: building an intelligent prompt library: defining a multi-model comparison and selection rule, a prompt template, a prompt template, a prompt parameter, a context data source, an external data source and a data format for each intelligent data prompt; transaction scheduling: importing the required intelligent prompt library into the transaction scheduling module, and submitting the intelligent prompts to the AI service one by one by the transaction scheduling module; the AI service is used for supporting a multi-modal large model for processing multi-source data; data summarization: supporting preprocessing and recording of return results of AI service processing each time in a transaction scheduling module; and document output: an output module reads the template and generates a target file in combination with the input data. According to the method, the problem that a large amount of time is needed to summarize, examine and sort numerous bidding documents and issue standard bid opening tables and bid evaluation reports in the traditional bid opening and evaluation process is solved, and the working efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of bidding and procurement, and in particular to an artificial intelligence-based bid evaluation method and system. Background Art

[0002] Bid evaluation refers to the act of the tenderer examining, reviewing and comparing each bid document according to the bid evaluation criteria and methods given in the tender documents. With the increase in public procurement activities and the increase in the number of enterprises participating in the bidding, the work of bid opening and bid evaluation has become more onerous. At the same time, due to the characteristics of some projects, the project bidding is highly centralized due to the procurement bidding sections. This trend further makes these tasks more onerous and complicated. There are some digital bid evaluation and intelligent assisted bid evaluation methods, which are assisted by machine learning algorithms, but they all require the initialization of the bid documents before the automated analysis of the bid data can be performed. Although this can improve the efficiency and accuracy of bid evaluation, it cannot help reduce the workload of bid opening and bid evaluation.

[0003] Traditional intelligent auxiliary tools such as bid clearing, bid opening and bid evaluation either use pre-developed programs to process data in standard formats, or use custom-developed artificial intelligence programs such as neural networks or NLP to analyze and interpret text from a probabilistic statistical perspective. The former relies on data in a fixed format, and the latter, although it can process text and other content, relies on the same type of text for processing and cannot process other flexible texts. Most importantly, neither can be flexibly expanded, and additional functions need to be customized and developed each time to be compatible with more types of procurement and bid evaluation tasks.

[0004] At present, there is no application case that combines prompt words and large models to form a business field. For intelligent auxiliary applications in the bid clearance and bid evaluation stages, traditional methods or NLP (natural language processing) solutions are often used. The traditional method is for personnel to organize data into a fixed format and then perform program judgment. If the text is reviewed, NLP methods such as neural networks are often used to process contract documents, which have high requirements for program development and input files. Once the text format changes greatly, such as text that exceeds the content of the training set, or involves a certain depth of business processing analysis, such as determining whether the registered capital meets the standard, the accuracy is not high, and it is difficult to flexibly expand and respond to different review materials. At present, some large models on the market have template functions similar to the intelligent prompts in this solution, but this template function can only realize the input of the current conversation, and cannot meet the data context communication in multiple interactions, resulting in its single function use and inability to realize some complex business logic analysis and calculations.

[0005] An AI-based bid evaluation method refers to the use of AI intelligent technologies such as large models, OCR, and NLP to assist in analyzing the bid documents received during the bidding process. While automatically processing and analyzing the bid content, it automatically conducts a preliminary review based on the preliminary review rules, provides a bid opening form, summarizes the business and technical scores after the bid opening, and conducts intelligent scoring and selection in combination with the bid evaluation rules to generate a bid evaluation report. This method is designed to assist procurement personnel in completing part of the bid opening and evaluation work, to address the strong dependence of transmission bid opening and evaluation auxiliary tools on file and data formats, the need for pre-development of review content, and the inability to flexibly respond to procurement reviews in different situations. It solves the traditional bid opening and evaluation process, which requires a lot of time to summarize, review, and organize numerous bid documents and issue standardized bid opening forms and bid evaluation reports, thereby improving work efficiency. Summary of the invention

[0006] The technical problem to be solved by the present invention is to provide an artificial intelligence-based bid evaluation method and system, which aims to assist procurement personnel in completing part of the bid opening and evaluation work, to cope with the strong dependence of transmission bid opening and evaluation auxiliary tools on file and data formats, the need for pre-development of review content, and the inability to flexibly respond to procurement reviews in different situations, and to solve the problem that the traditional bid opening and evaluation process requires a lot of time to summarize, review and organize numerous bidding documents and issue standardized bid opening forms and evaluation reports, thereby improving work efficiency.

[0007] In order to solve the above technical problems, the technical solution adopted by the present invention is: A bid evaluation method based on artificial intelligence comprises the following steps: S1. Establish an intelligent prompt library: define multi-model comparison rules, prompt templates, prompt templates, prompt parameters, context data sources, external data sources and data formats for each intelligent data prompt; form complete prompt content through prompt templates, prompt templates and prompt parameters to guide the big model to perform business processing of the current intelligent prompt transaction, including information extraction, indicator calculation and logical judgment functions; the data source required for the transaction includes the data content that needs to be provided to the big model in this transaction; when the big model is required to output the business processing results in a specified format, define the format of the returned data; S2. Transaction Scheduling: Import the required intelligent prompt library into the transaction scheduling module, set the execution order of each intelligent prompt data, initialize the relevant content, and let the transaction scheduling module submit the intelligent prompts to the AI ​​service one by one. After listening to the results returned by AI processing, format the data according to the result conversion and record it in the context of this scheduling for subsequent data aggregation; S3, AI service: supports multimodal large models for processing multi-source data, including text and images; also supports text generation, OCR recognition, logical reasoning and analysis control, and combines different intelligent prompts into intelligent prompts to meet the automation processing functions of each link in bid evaluation; S4, Data aggregation: supports the preprocessing and recording of the results returned by each AI service in the transaction scheduling module. On the one hand, it is used as input in subsequent intelligent prompts, and on the other hand, it is finally transmitted to the document output module as context file data and used as data input for document output. S5. Document output: After pre-defining the Excel or Word template, a fixed-format data interface is given. When the transaction scheduling module calls the document output and inputs data in the corresponding format, the document output module reads the template and generates a target file based on the input data. The target file includes the bid opening form and bid evaluation report.

[0008] A further improvement of the technical solution of the present invention is that: in S1, it also includes: after the prompt template is set, there will be context data and external data contained in the prompt of the specified smart prompt, which are used for the transaction scheduling module to read and insert the data into the prompt.

[0009] A further improvement of the technical solution of the present invention is that the intelligent prompting service includes summarizing and organizing the extracted information and the review information, and summarizing them again according to the data required by the bid opening form to generate the bid opening form.

[0010] A further improvement of the technical solution of the present invention is that the intelligent prompt transaction includes an intelligent preliminary review of the information extracted from the bidding documents based on the requirements of the preliminary review, and directly conducts a preliminary review of the bidding documents for two-way verification. The preliminary review content includes: bidding document content, bidding format, price amount, minimum / maximum price review, credit qualification, financial status and collusion judgment.

[0011] A further improvement of the technical solution of the present invention is that after passing the qualification review and preliminary evaluation, the bidding documents are verified and judged according to the business review terms. The business review content includes the response to the requirements of the business documents, the judgment of the enterprise qualifications, the judgment of the financial status and social reputation, and the judgment of after-sales service.

[0012] A further improvement of the technical solution of the present invention is that: in S2, it also includes: the transaction scheduling module forms a complete prompt content by combining the prompt template, prompt template and prompt parameters of an intelligent prompt data. If the intelligent prompt data contains external data or data in the context, the transaction scheduling module attaches these additional data when sending the prompt.

[0013] A further improvement of the technical solution of the present invention is that in S3, a method is adopted to guide the AI ​​service to calculate sub-item scores based on price, business and technology, and then calculate a comprehensive scoring method to make manufacturer recommendations based on the comprehensive scores.

[0014] A further improvement of the technical solution of the present invention is that in S4, the AI ​​is guided to provide the data required for the review report through preset prompt content, which is used to generate data input for the review report.

[0015] An artificial intelligence-based bid evaluation system comprises an intelligent prompt database, a transaction scheduling module, an AI service module, a data aggregation module and a document output module; the intelligent prompt database, the transaction scheduling module and the document output module are connected in sequence; the AI ​​service module and the data aggregation module are interactively connected with the transaction scheduling module respectively; Smart prompt library: used as an initialization transaction when processing bid evaluation transaction scheduling, including several smart prompt data; each smart prompt data includes multi-model comparison rules, prompt template, prompt template, prompt parameter, context data source, external data source and data format; Transaction scheduling module: On the one hand, it supports users to organize the smart prompts in one or more smart prompt libraries in a certain order, and after setting the preset parameters and content of each smart prompt, the module will execute the smart prompt according to the set execution rules; on the other hand, it supports the abnormal information processing and fault recovery during transaction processing; AI service module: It uses common network protocols to provide text generation, OCR recognition, logical reasoning and analysis control based on the received text or pictures. It can process multimodal large models of data such as text and images, and combines different intelligent prompts to meet the automation processing functions of each link in the bid evaluation. Data aggregation module: On the one hand, it pre-processes the data and stores and records them; on the other hand, it combines the data with intelligent prompts as the input of AI services for secondary processing and in-depth business analysis of the data. Finally, all the data is transferred to the document output module through transaction scheduling as the data input for the standard document bid opening form and bid evaluation report; Document output module: receives predefined data, fills the data into a preset fixed Excel or Word template, and finally forms an Excel / Word document, which is commonly used as a bid opening or bid evaluation result file.

[0016] Due to the adoption of the above technical solution, the technical progress achieved by the present invention is: The present invention combines a flexible intelligent prompt library with a multi-source large model. The multi-source large model can process texts and pictures, extract information, generate text, perform logical reasoning, and perform computational analysis. Flexible prompt words are used to guide the multi-source large model to perform bid evaluation. This can not only improve the accuracy of AI processing, but also flexibly expand the intelligent prompt library without the need for secondary development, so that the system can have the flexibility and reliability to process more different contract bidding documents for bid evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic diagram of the system structure of the present invention; Figure 2 It is a schematic diagram of the improved Prompt solution of the present invention; Figure 3 It is a schematic diagram of generating a bid evaluation report of the present invention. DETAILED DESCRIPTION

[0018] The present invention is further described in detail below with reference to the accompanying drawings and embodiments: like Figure 1 As shown, an artificial intelligence-based bid evaluation system includes an intelligent prompt database, a transaction scheduling module, an AI service module, a data aggregation module and a document output module; the intelligent prompt database, the transaction scheduling module and the document output module are connected in sequence; the AI ​​service module and the data aggregation module are interactively connected with the transaction scheduling module respectively; Smart prompt library: used as an initialization transaction when processing bid evaluation transaction scheduling, including several smart prompt data; each smart prompt data includes multi-model comparison rules, prompt template, prompt template, prompt parameter, context data source, external data source and data format; Transaction scheduling module: On the one hand, it supports users to organize the smart prompts in one or more smart prompt libraries in a certain order, and after setting the preset parameters and content of each smart prompt, the module will execute the smart prompt according to the set execution rules; on the other hand, it supports the abnormal information processing and fault recovery during transaction processing; AI service module: It uses common network protocols to provide text generation, OCR recognition, logical reasoning and analysis control based on the received text or pictures. It can process multimodal large models of data such as text and images, and combines different intelligent prompts to meet the automation processing functions of each link in the bid evaluation. Data aggregation module: On the one hand, it pre-processes the data and stores and records them; on the other hand, it combines the data with intelligent prompts as the input of AI services for secondary processing and in-depth business analysis of the data. Finally, all the data is transferred to the document output module through transaction scheduling as the data input for the standard document bid opening form and bid evaluation report; Document output module: receives predefined data, fills the data into a preset fixed Excel or Word template, and finally forms an Excel / Word document, which is commonly used as a bid opening or bid evaluation result file.

[0019] A bid evaluation method based on artificial intelligence comprises the following steps: S1. Establish an intelligent prompt library: define multi-model comparison rules, prompt templates, prompt templates, prompt parameters, context data sources, external data sources and data formats for each intelligent data prompt; form complete prompt content through prompt templates, prompt templates and prompt parameters to guide the big model to perform business processing of the current intelligent prompt transaction, including information extraction, indicator calculation and logical judgment functions; the data source required for the transaction includes the data content that needs to be provided to the big model in this transaction; when the big model is required to output the business processing results in a specified format, define the format of the returned data; S1 also includes: when the prompt template is set, context data and external data contained in the prompt of the specified intelligent prompt will be set, which is used for the transaction scheduling module to read and insert the data into the prompt; The intelligent prompting service includes summarizing and arranging the extracted information and the review information, and summarizing them again according to the data required by the bid opening form to generate the bid opening form; The intelligent prompt transaction includes an evaluation method, which can be expanded into one or more steps, and can guide the large model to conduct special scoring and selection for each manufacturer according to requirements.

[0020] Intelligent prompting services include intelligent preliminary review of the information extracted from the bidding documents based on the requirements of the preliminary review. It can also directly conduct qualification review and preliminary review of the bidding documents at the same time, and conduct two-way verification. The preliminary review content includes: bidding document content, bidding format, price amount, minimum / maximum price review, credit qualification, financial status, collusion judgment, etc.

[0021] After passing the qualification review and preliminary evaluation, the bidding documents are verified and judged according to the business review terms. The business review content includes the response to the requirements of the business documents (negotiation validity period, delivery cycle, power of attorney, response to contract terms), the judgment of enterprise qualifications (connecting to the official website database, approving the manufacturer's business license, organization code, tax registration certificate, general tax certificate, production license, qualification certificate and quality management certification), the judgment of financial status and social reputation (verifying the registered capital of the enterprise in the bidding documents online, verifying whether the bidding documents have a balance sheet, income statement, cash flow statement and audit report, and verifying whether the bidding documents contain materials such as the enterprise load rate and bank credit certificate), and after-sales service judgment (verifying the after-sales service and commitments, technical service plans in the bidding documents, and extracting user comprehensive evaluation summary information from the user comprehensive evaluation in the bidding documents).

[0022] S2. Transaction Scheduling: Import the required intelligent prompt library into the transaction scheduling module, set the execution order of each intelligent prompt data, initialize the relevant content, and let the transaction scheduling module submit the intelligent prompts to the AI ​​service one by one. After listening to the results returned by AI processing, format the data according to the result conversion and record it in the context of this scheduling for subsequent data aggregation. At the same time, the transaction scheduling module handles the exception processing of each transaction result and supports fault recovery, which improves the robustness and ease of use of the system. S2 also includes: the transaction scheduling module forms a complete prompt content by combining a prompt template, a prompt template and prompt parameters of a piece of intelligent prompt data. If the intelligent prompt data contains external data or data in the context, the transaction scheduling module will attach these additional data when sending the prompt; a reference example is to extract information from the bidding documents through the intelligent prompt transaction, such as legal person information, client information, business license, registered capital, bank credit, quotation information and other key information in the business contract content; S3, AI service: supports multimodal large models for processing multi-source data, including text and images; also supports text generation, OCR recognition, logical reasoning and analysis control, and combines different intelligent prompts into intelligent prompts to meet the automation processing functions of each link in bid evaluation; In this case, the AI ​​service was guided to calculate sub-scores based on price, business, and technology, and then a comprehensive scoring method was calculated to make manufacturer recommendations based on the comprehensive score.

[0023] Through the preset prompt content, AI is guided to provide the data required for the review report, which is used to generate data input for the review report.

[0024] S4. Data aggregation: supports the preprocessing and recording of the return results of each AI service processing in the transaction scheduling module. On the one hand, it is used as input in subsequent intelligent prompts, and on the other hand, it is finally transmitted to the document output module as context file data and used as data input for document output.

[0025] S5. Document output: After pre-defining the Excel or Word template, a fixed-format data interface is given. When the transaction scheduling module calls the document output and inputs data in the corresponding format, the document output module reads the template and generates a target file based on the input data. The target file includes the bid opening form and bid evaluation report. Example

[0026] like Figure 2 As shown, an improvement on an AI-based bid evaluation method using Prompt enables AI to focus on current business transaction processing.

[0027] The transaction scheduling module generates a complete prompt and submits it to the AI ​​service based on the prompt template, external data source and transaction context of the smart prompt, the execution instructions and data source of the business transaction, and the prompt template and prompt content of the smart prompt.

[0028] When the transaction scheduling module receives the content returned by the AI ​​service, it formats the result according to the data format of the intelligent prompt, transmits the result to the data summary module, and records it as the data in the current transaction context, thereby enhancing the AI ​​service to focus more on the current prompt instruction when processing different business logics and reducing the number of business-irrelevant tokens.

[0029] like Figure 3 As shown, the system execution steps of an artificial intelligence-based bid evaluation method under a specific bid evaluation system are as follows: Step 1. According to the preset intelligent prompt execution sequence, the transaction scheduling loads the bidding document processing intelligent prompt set. First, the template of the intelligent prompt is loaded, such as guiding the AI ​​service to extract text information and business license and other image information. Then, combined with the instruction content of this intelligent prompt, the instruction content includes guiding the AI ​​to extract information from each manufacturer's bidding documents, what information to extract, and the final record return format of the information; the transaction scheduling module sends the prompts in the bidding document intelligent processing prompt set to the AI ​​service one by one. After receiving the intelligent prompt instruction for bidding document processing information, the AI ​​service generates text and extracts information one by one, and returns the content to the transaction scheduling in the format required by the instruction. The transaction scheduling determines whether an error or failure occurs according to the setting of the intelligent prompt and records the returned information.

[0030] Step 2: The transaction scheduler loads the preliminary review smart prompt set, which contains a series of review lists for the manufacturer's bidding documents and content. Each review list corresponds to one or more smart prompt sets, and each smart prompt set contains one or more smart prompts to guide the AI ​​service to review the manufacturer's bidding documents and the currently extracted information. The transaction scheduler module sends the smart prompts to the AI ​​service one by one in the preset order. After receiving the smart prompt instructions for preliminary review, the AI ​​service performs text retrieval and review judgment, and returns the review results to the transaction scheduler according to the instructions.

[0031] The transaction scheduler determines whether an error or failure occurs based on the settings of the intelligent prompt and records the returned information.

[0032] Step 3: The transaction scheduling loads the bid opening data generation instruction set, which contains instructions to guide the AI ​​service to combine the extracted information with the review content to generate the next candidate manufacturer and review information as the data input for generating the bid opening form. The transaction scheduling module calls the document output module and passes the input data of the bid opening form. The document output module loads the specified template and generates the bid opening form according to the format and content agreed in the template. The form contains the preliminary review results, business review form, and technical review form.

[0033] Step 4: The transaction scheduler loads the scoring and selection instruction set, which includes: 1. Instructions for parsing business scores and technical scores, which are used to guide the AI ​​service to extract the judges' scoring data for each manufacturer; 2. Instructions for parsing manufacturer quotation files, which guide the AI ​​service to extract the quotations of each manufacturer and perform price scoring; 3. Comprehensive score calculation instructions, which guide the AI ​​service to calculate the final score of each manufacturer according to the price, business and technical scores and the comprehensive score calculation logic; 4. Recommended sorting instructions, which guide the AI ​​to sort the recommended order of manufacturers according to certain rules. The transaction scheduling module sends these instructions to the AI ​​service one by one. After receiving the intelligent prompt instructions, the AI ​​service performs business calculations and logical reasoning, and returns the results to the transaction scheduler according to the instruction requirements. The transaction scheduler determines whether an error or failure occurs based on the settings of the intelligent prompt, and records the returned information. Step 5: The transaction scheduling module loads the instructions for generating the bid evaluation report. The instruction set includes: guiding AI to summarize the extracted information, review results, price score calculation, technical score calculation, business score calculation and other information, and output them in the specified format. The transaction scheduling module sends these instructions to the AI ​​service one by one. After receiving the intelligent prompt instructions, the AI ​​service performs business calculations and logical reasoning, and returns the results to the transaction scheduling according to the instructions. After the transaction scheduling determines that the execution is successful based on the returned results, it calls the document output module, and the input data is the result returned by the AI ​​service. The document output module loads the specified template and generates a bid evaluation report with the input data in the format and content agreed upon by the template.

Claims

1. A bid evaluation method based on artificial intelligence, characterized in that: The following steps are involved: S1. Establish an intelligent prompt library: define multi-model comparison rules, prompt templates, prompt templates, prompt parameters, context data sources, external data sources, and data formats for each intelligent data prompt; The prompt template, prompt template and prompt parameters form a complete prompt content to guide the big model to perform business processing of the current intelligent prompt transaction. The business includes information extraction, indicator calculation and logical judgment functions. The data source required for the transaction includes the data content that needs to be provided to the big model in this transaction. When the big model is required to output the business processing results in a specified format, the format of the returned data is defined. S2. Transaction Scheduling: Import the required intelligent prompt library into the transaction scheduling module, set the execution order of each intelligent prompt data, initialize the relevant content, and let the transaction scheduling module submit the intelligent prompts to the AI ​​service one by one. After listening to the results returned by AI processing, format the data according to the result conversion and record it in the context of this scheduling for subsequent data aggregation; S3, AI service: supports multimodal large models for processing multi-source data, including text and images; also supports text generation, OCR recognition, logical reasoning and analysis control, and combines different intelligent prompts into intelligent prompts to meet the automation processing functions of each link in bid evaluation; S4, Data aggregation: supports the preprocessing and recording of the results returned by each AI service in the transaction scheduling module. On the one hand, it is used as input in subsequent intelligent prompts, and on the other hand, it is finally transmitted to the document output module as context file data and used as data input for document output. S5. Document output: After pre-defining the Excel or Word template, a fixed-format data interface is given. When the transaction scheduling module calls the document output and inputs data in the corresponding format, the document output module reads the template and generates a target file based on the input data. The target file includes the bid opening form and bid evaluation report.

2. The bid evaluation method based on artificial intelligence according to claim 1, characterized in that: S1 also includes: when the prompt template is set, context data and external data contained in the prompt of the specified smart prompt will be set, which is used for the transaction scheduling module to read and insert the data into the prompt.

3. The bid evaluation method based on artificial intelligence according to claim 1, characterized in that: The intelligent prompting service includes summarizing and organizing the extracted information and the review information, and summarizing them again according to the data required by the bid opening form to generate the bid opening form.

4. The bid evaluation method based on artificial intelligence according to claim 1, characterized in that: The intelligent prompt service includes intelligent preliminary review of the information extracted from the bidding documents based on the requirements of the preliminary review, and directly conducting qualification review and preliminary review of the bidding documents, and conducting two-way verification. The preliminary review content includes: bidding document content, bidding format, price amount, minimum / maximum price review, credit qualification, financial status, data calculation errors and collusion judgment.

5. The bid evaluation method based on artificial intelligence according to claim 4 is characterized in that: After passing the qualification review and preliminary evaluation, the bidding documents will be verified and judged according to the business review terms. The business review content includes the response to the requirements of the business documents, the judgment of the enterprise qualifications, the judgment of the financial status and social reputation, and the judgment of after-sales service.

6. The bid evaluation method based on artificial intelligence according to claim 1, characterized in that: S2 also includes: the transaction scheduling module forms a complete prompt content by combining a prompt template, a prompt template and prompt parameters of a smart prompt data. If the smart prompt data contains external data or data in the context, the transaction scheduling module attaches these additional data when sending the prompt.

7. The bid evaluation method based on artificial intelligence according to claim 1, characterized in that: In S3, the AI ​​service is guided to calculate sub-item scores based on price, business and technology, and then calculate the comprehensive scoring method to make manufacturer recommendations based on the comprehensive score.

8. The bid evaluation method based on artificial intelligence according to claim 1, characterized in that: In S4, the preset prompt content guides AI to provide the data required for the review report, which is used to generate data input for the review report.

9. An artificial intelligence-based bid evaluation system, characterized by: Using the artificial intelligence-based bid evaluation method as described in any one of claims 1 to 8, comprising an intelligent prompt database, a transaction scheduling module, an AI service module, a data aggregation module and a document output module; the intelligent prompt database, the transaction scheduling module and the document output module are connected in sequence; the AI ​​service module and the data aggregation module are interactively connected with the transaction scheduling module respectively; Intelligent prompt library: used as the initialization task when processing the bidding evaluation task scheduling, including several pieces of intelligent prompt data; Each intelligent prompt data content includes multi-model comparison rules, prompt template, prompt template, prompt parameters, context data source, external data source and data format; Transaction scheduling module: On the one hand, it supports users to organize the smart prompts in one or more smart prompt libraries in a certain order, and after setting the preset parameters and content of each smart prompt, the module will execute the smart prompt according to the set execution rules; on the other hand, it supports the abnormal information processing and fault recovery during transaction processing; AI service module: It uses common network protocols to provide text generation, OCR recognition, logical reasoning and analysis control based on the received text or pictures. It can process multimodal large models of data such as text and images, and combines different intelligent prompts to meet the automation processing functions of each link in the bid evaluation. Data aggregation module: On the one hand, it pre-processes the data and stores and records them; on the other hand, it combines the data with intelligent prompts as the input of AI services for secondary processing and in-depth business analysis of the data. Finally, all the data is transferred to the document output module through transaction scheduling as the data input for the standard document bid opening form and bid evaluation report; Document output module: receives predefined data, fills the data into a preset fixed Excel or Word template, and finally forms an Excel / Word document, which is commonly used as a bid opening or bid evaluation result file.

Images