A method for overall architecture design of an intelligent bidding assistance system

Abstract

The application discloses a kind of overall architecture design methods of intelligent tendering auxiliary system, belong to tendering auxiliary technical field, including S1. Architecture requirement hierarchical analysis;S2. Intelligent core layer module design;S3. Data support layer architecture design;S4. Interface adaptation layer and interactive layer collaborative design;S5. Security protection layer embedded design;S6. Architecture linkage debugging and iterative optimization;Five-layer architecture hierarchical design is adopted, the function orientation and requirement of each layer architecture are clear;The overall architecture design method of the intelligent tendering auxiliary system establishes the corresponding relationship of each layer requirement and module by requirement-module correlation matrix, solves the problem of existing architecture design fragmentation, realizes the close linkage between each layer architecture and each module, ensures the seamless connection and efficient operation of tendering whole process.

Description

Technical Field

[0001] This invention belongs to the field of bidding assistance technology, specifically relating to an overall architecture design method for an intelligent bidding assistance system. Background Technology

[0002] With the standardization and digitalization of the bidding industry, the frequency of corporate bidding activities is constantly increasing, and the complexity and professionalism of the bidding process are also increasing. The requirements for the intelligence, efficiency and integration of bidding support systems are getting higher and higher. The architecture design of existing bidding support systems often suffers from fragmentation, with each functional module being independent and having poor linkage, making it impossible to achieve seamless connection and intelligent support for the entire bidding process.

[0003] In existing technologies, the parsing of bidding data largely relies on manual operation, which is inefficient and prone to errors; the formulation of bidding strategies lacks intelligent algorithm support and relies heavily on experience-based judgment, resulting in poor adaptability; data storage uses a single database model, which cannot meet the differentiated storage needs of structured and unstructured data, and the timeliness of data retrieval is insufficient; the interface adaptability is weak, making it difficult to be compatible with the communication protocols of different external bidding platforms, resulting in low data interaction efficiency; security protection mostly adopts a single protection mode, which cannot comprehensively guarantee the security of bidding data and the standardization of operations; at the same time, after the architecture design is completed, there is a lack of effective linkage debugging and iterative optimization mechanisms, making it difficult to adapt to constantly changing bidding business needs, resulting in a significant reduction in system usability and failing to effectively improve the efficiency and success rate of enterprise bidding. Summary of the Invention

[0004] The purpose of this invention is to provide an overall architecture design method for an intelligent bidding assistance system to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for designing the overall architecture of an intelligent bidding assistance system, comprising the following steps: S1. Layered Analysis of Architectural Requirements: Based on the entire bidding process, the basic architectural requirements are broken down into five layers: user interaction layer, intelligent core layer, data support layer, interface adaptation layer, and security protection layer. Specifically, the user interaction layer corresponds to the bidding entity's operational scenarios; the intelligent core layer corresponds to bidding data analysis, strategy generation, and risk warning scenarios; the data support layer corresponds to bidding data storage, retrieval, and update scenarios; the interface adaptation layer corresponds to data interaction scenarios with external bidding platforms and internal enterprise management systems; and the security protection layer corresponds to bidding data confidentiality and operational permission control scenarios. Furthermore, a requirement priority ranking algorithm is used to filter the core requirements of each layer, establish a mapping relationship between requirements and architectural modules, and generate a requirement-module association matrix to provide a basis for subsequent architecture design. S2. Intelligent Core Layer Module Design: Based on the core requirements of the intelligent core layer obtained in step S1, an intelligent core layer is constructed, consisting of a bid data intelligent parsing module, a bid strategy adaptive generation module, a bid risk real-time early warning module, and a multi-dimensional comparative analysis module. The bid data intelligent parsing module is equipped with an OCR recognition and NLP semantic analysis linkage algorithm to parse key parameters, qualification requirements, and scoring criteria in bidding announcements and bidding documents. The bid strategy adaptive generation module is linked with the bid data intelligent parsing module, generating personalized bid strategies adapted to the current bidding project based on the parsed key information, combined with historical bidding data and industry benchmark data. The bid risk real-time early warning module is linked with both the bid data intelligent parsing module and the bid strategy adaptive generation module, providing tiered early warnings for qualification mismatches, parameter omissions, and unreasonable strategies discovered during the parsing process. S3. Data Support Layer Architecture Design: Based on the data support layer requirements in step S1 and the data call requirements of each module in the intelligent core layer in step S2, a distributed data storage architecture is constructed. This distributed data storage architecture includes a structured database, an unstructured database, and a real-time cache database. The structured database stores structured data such as the qualifications of bidding entities and historical bidding records. The unstructured database stores unstructured data such as bidding documents, tender documents, and scanned copies of qualification certificates. The real-time cache database caches temporary data and frequently called data during the parsing and calculation process of the intelligent core layer. Simultaneously, a data lifecycle management module is designed to automate the entire process of data collection, cleaning, desensitization, storage, updating, and destruction. The data collection stage is linked with the interface adaptation layer, and the data cleaning and desensitization stages provide high-quality data support for the intelligent core layer. S4. Collaborative Design of Interface Adaptation Layer and Interaction Layer: Based on the requirements of the interface adaptation layer and user interaction layer in step S1, and combined with the module functions of the intelligent core layer and data support layer in steps S2 and S3, a multi-protocol adaptive interface module for the interface adaptation layer is designed. This multi-protocol adaptive interface module supports multiple protocols such as HTTP, HTTPS, and WebService, enabling seamless data interaction with external bidding platforms and internal enterprise management systems. It also works in conjunction with the data acquisition module of the data support layer to automatically capture bidding data and automatically upload bid data. Simultaneously, a multi-terminal adaptive interface for the user interaction layer is designed. This interface works in conjunction with the modules of the intelligent core layer to realize the visualization of bid data, strategy editing, risk viewing, and operation command issuance, supporting seamless switching between PC, mobile, and tablet terminals. S5. Embedded Design of Security Protection Layer: Based on the security protection layer requirements in step S1, and combined with the functional characteristics of each layer architecture in steps S2-S4, the security protection module is embedded and integrated into each layer architecture. The security protection module includes a permission hierarchical control module, a data encryption transmission module, an operation log auditing module, and an abnormal behavior interception module. Among them, the permission hierarchical control module is linked with the user interaction layer to assign different operation permissions based on the bidding role; the data encryption transmission module is linked with the interface adaptation layer and the data support layer to realize end-to-end encryption during data transmission and storage; the operation log auditing module is linked with each layer module to record all operation behaviors for traceability and verification; the abnormal behavior interception module is linked with the intelligent core layer to intercept malicious operations and abnormal data access based on behavioral feature analysis. S6. Architecture Linkage Debugging and Iterative Optimization: Integrate the architecture modules designed in steps S2-S5 to construct a complete intelligent bidding assistance system architecture; debug the linkage performance between the architecture layers by simulating the entire bidding process, verifying the stability of interface adaptation, the timeliness of data calls, the accuracy of intelligent strategies, and the reliability of security protection; based on the debugging results, combined with the requirement-module association matrix in step S1, iteratively optimize the functional parameters of each module to form a feasible and highly adaptable intelligent bidding assistance system architecture. The iterative optimization process is linked with the data analysis module of the intelligent core layer, automatically generating optimization suggestions based on debugging data and historical operating data.

[0006] As a preferred implementation method, the requirement priority ranking algorithm in step S1 is as follows: a three-dimensional evaluation system is constructed based on the impact of bidding business, the urgency of user needs and the difficulty of technical implementation. The comprehensive priority score of each requirement is calculated by the analytic hierarchy process, and the requirement with a comprehensive score ≥ 80 is selected as the core requirement. In the three-dimensional evaluation system, the weight of business impact is 0.4, the weight of requirement urgency is 0.3, and the weight of technical implementation difficulty is 0.3.

[0007] As a preferred implementation, the OCR recognition and NLP semantic analysis linkage algorithm in step S2 specifically includes: extracting text information from text, tables, and images in the bidding documents through the OCR recognition module to generate standardized text; the NLP semantic analysis module performing word segmentation, part-of-speech tagging, and entity recognition on the standardized text to extract key information such as the bidding project name, bidding unit, qualification requirements, scoring rules, and bid deadline, and establishing a key information tag library. The key information tag library is linked with the structured database of the data support layer to realize rapid retrieval and access of key information.

[0008] As a preferred implementation, the personalized bidding strategy described in step S2 is specifically generated as follows: the adaptive bidding strategy generation module calls historical successful bidding case data and industry benchmark data in the data support layer, compares them with the key information of the current bidding project, constructs a strategy generation model through machine learning algorithms, and outputs qualification matching suggestions, price range suggestions, technical solution optimization suggestions, and key suggestions for bid document preparation. The parameters of the strategy generation model are continuously trained and optimized through historical bidding data, and are linked with the real-time bidding risk warning module to perform risk verification on the generated bidding strategy.

[0009] As a preferred implementation, the data lifecycle management module in step S3 specifically performs the following data cleaning process: deduplication, noise reduction, missing value filling, and format standardization on the collected bidding-related data. The missing value filling adopts a fusion filling algorithm based on industry average and historical similar data. The data desensitization process specifically involves encrypting and desensitizing the core qualification information and quotation information of the bidding entity using an irreversible encryption algorithm to ensure that the desensitized data cannot be restored while retaining the statistical analysis value of the data.

[0010] As a preferred implementation, the multi-protocol adaptive interface module in step S4 has the function of automatic protocol identification and switching. Specifically, it detects the communication protocol of the external interface system through the interface detection module, automatically matches the corresponding interface protocol, and completes the adaptive adjustment of data interaction parameters. If an unknown protocol is detected, the protocol parsing module is automatically triggered to parse based on the protocol feature library and generate a temporary adaptive interface to ensure the continuity of data interaction. The protocol feature library is linked with the data support layer to achieve real-time updates.

[0011] As a preferred implementation, the abnormal behavior interception module in step S5 has the following behavior feature analysis process: based on the historical operation data of the intelligent core layer, a normal operation behavior feature model is constructed, the current operation behavior data is collected in real time and compared with the normal behavior feature model. When the comparison deviation exceeds a preset threshold, it is determined to be abnormal behavior, an interception command is immediately triggered, and a warning message is sent to the administrator through the real-time warning module for bidding risks. The preset threshold can be adaptively adjusted according to the actual application scenario.

[0012] As a preferred implementation, the iterative optimization in step S6 specifically includes: based on the data call time, strategy generation accuracy, risk warning accuracy, and user operation feedback data during system operation, multi-dimensional statistical analysis is performed through the data analysis module of the intelligent core layer to generate optimization reports for each layer module. The optimization reports include module functional defects, parameter adjustment suggestions, and performance improvement schemes. Targeted adjustments are made to each layer module according to the optimization reports to form a closed-loop process of "design-debugging-optimization-iteration".

[0013] As a preferred implementation, the multi-terminal adaptation interface in step S4 also integrates a bidding progress visualization tracking module. The tracking module is linked with the intelligent core layer and the data support layer to capture the progress data of each stage of the bidding process (document preparation, qualification review, quotation confirmation, submission and upload) in real time and display it in the form of a timeline. At the same time, progress node warnings are set. When a node is not completed within the time limit, the real-time bidding risk warning module is triggered to send a reminder message.

[0014] As a preferred implementation, the real-time cache database in step S3 adopts a cache eviction strategy. The cache eviction strategy is constructed based on data call frequency, data timeliness, and data importance. It automatically evicts temporary data that has a call frequency lower than a preset threshold, has exceeded its validity period, or is non-core, ensuring the storage efficiency of the cache database and the timeliness of data calls. The preset threshold is linked with the data analysis module of the intelligent core layer and is adaptively adjusted according to historical call data.

[0015] Compared with the prior art, the beneficial effects of the present invention are: The overall architecture design of this intelligent bidding assistance system adopts a five-layer architecture, which clarifies the functional positioning and requirements of each layer. By establishing the correspondence between the requirements and modules of each layer through a requirement-module association matrix, the fragmentation problem of the existing architecture design is solved, and the close linkage between each layer and each module is achieved, ensuring seamless connection and efficient operation of the entire bidding process. The overall architecture design of this intelligent bidding assistance system integrates multi-module linkage design in the intelligent core layer and is equipped with intelligent technologies such as OCR recognition and NLP semantic analysis linkage algorithm and machine learning strategy generation model. It realizes automatic parsing of bidding data, personalized generation of bidding strategies, and real-time warning of bidding risks, replacing traditional manual operation, greatly improving bidding efficiency and intelligence level, reducing bidding risks, and increasing the success rate of bidding. The overall architecture design method of this intelligent bidding assistance system constructs a distributed data storage architecture and a data lifecycle management module to achieve differentiated storage of structured and unstructured data and high-quality management of the entire data process. Combined with the cache eviction strategy of the real-time cache database, it improves the timeliness of data retrieval and storage efficiency, and provides reliable data support for the intelligent core layer.

[0016] The overall architecture design method of this intelligent bidding assistance system includes the design of multi-protocol adaptive interface modules and multi-terminal adaptable interfaces, which improves interface compatibility and user interaction convenience, realizes seamless data interaction with external systems and flexible operation on multiple terminals, and adapts to the operational needs of different bidding entities.

[0017] The overall architecture design of this intelligent bidding assistance system adopts an embedded security protection layer design, integrating security protection modules into each layer of the architecture to achieve comprehensive security protection including access control, data encryption, log auditing, and anomaly interception, ensuring the security of bidding data and system operations.

[0018] The overall architecture design method of this intelligent bidding assistance system establishes a closed-loop process of "design-debugging-optimization-iteration". Through coordinated debugging and iterative optimization, it ensures that the architecture always adapts to the ever-changing bidding business needs, improves the system's adaptability, stability and long-term usability, and has good promotional value. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the steps of the present invention; Detailed Implementation

[0020] The present invention will be further described below with reference to embodiments.

[0021] The following embodiments are used to illustrate the present invention, but should not be used to limit the scope of protection of the present invention. The conditions in the embodiments can be further adjusted according to specific conditions, and simple improvements to the method of the present invention under the premise of the concept of the present invention are all within the scope of protection claimed by the present invention.

[0022] Please see Figure 1 This invention provides an overall architecture design method for an intelligent bidding assistance system, comprising the following steps: S1. Layered Architecture Requirements Analysis: This section analyzes the entire bidding process, including stages such as obtaining bidding information, parsing bidding documents, formulating bidding strategies, preparing bid documents, submitting bid documents, tracking bidding progress, and controlling bidding risks. Based on these scenarios, a five-layer architecture is derived: a user interaction layer, an intelligent core layer, a data support layer, an interface adaptation layer, and a security protection layer. Specifically, the user interaction layer includes the bidding entity's user interface, data visualization, and command issuance; the intelligent core layer includes bidding document parsing, bidding strategy generation, bidding risk warning, and multi-dimensional comparative analysis; the data support layer includes bid data storage, data retrieval, data updates, and data cleaning and anonymization; the interface adaptation layer includes data interaction with external bidding platforms (such as the China Bidding and Tendering Public Service Platform), internal ERP systems, and qualification management systems; and the security protection layer includes access control, bid data encryption, operation log traceability, and abnormal behavior interception.

[0023] Simultaneously, a requirement priority ranking algorithm is used to screen core requirements: a three-dimensional evaluation system is constructed based on the impact of bidding business, the urgency of user needs, and the difficulty of technical implementation, with a weight of 0.4 for business impact, 0.3 for requirement urgency, and 0.3 for technical implementation difficulty. The comprehensive priority score of each requirement is calculated using the analytic hierarchy process, and requirements with a comprehensive score ≥80 are selected as core requirements, such as tender document parsing, bidding strategy generation, data encryption storage, and external platform interface adaptation. A mapping relationship between requirements and architecture modules is established, generating a requirement-module association matrix, clarifying the architecture modules corresponding to each core requirement, and providing a basis for subsequent module design.

[0024] S2. Intelligent Core Layer Module Design: Based on the core requirements of the intelligent core layer in step S1, a bidding data intelligent parsing module, a bidding strategy adaptive generation module, a bidding risk real-time early warning module, and a multi-dimensional comparative analysis module are constructed, and each module operates in conjunction with the others.

[0025] The intelligent bidding data analysis module is equipped with a combined OCR recognition and NLP semantic analysis algorithm. The OCR recognition module uses high-definition scanning recognition technology to extract text information from text, tables, and images in the bidding documents, remove redundant formatting, and generate standardized text. The NLP semantic analysis module uses the BERT word segmentation model to perform word segmentation, part-of-speech tagging, and entity recognition on the standardized text, extracting key information such as the bidding project name, bidding unit, qualification requirements (such as enterprise qualification level and personnel qualifications), scoring criteria (such as the proportion of technical and commercial parts), bid deadline, and price range. It establishes a key information tag library and stores the key information in a structured database of the data support layer, enabling rapid retrieval and access to key information.

[0026] The adaptive bidding strategy generation module works in conjunction with the intelligent bidding data analysis module. It calls upon historical successful bidding case data (similar projects in the same industry over the past 3 years) and industry benchmark data (such as industry average prices and qualification requirement benchmarks) from the data support layer. This data is compared with the key information of the current bidding project. A random forest algorithm is used to build a strategy generation model, which outputs qualification matching suggestions (such as the matching degree between the company's existing qualifications and the bidding requirements, and the qualification materials that need to be supplemented), price range suggestions (based on historical prices and industry benchmarks, combined with project cost calculations), technical solution optimization suggestions (such as technical parameter adjustments and case supplements), and key suggestions for bid document preparation (such as highlighting key scoring items). The parameters of the strategy generation model are continuously trained and optimized through historical bidding data. The model parameters are updated every 10 new bidding cases. At the same time, it is linked with the real-time bidding risk warning module to perform risk verification on the generated bidding strategy and identify risk points such as qualification mismatch, excessively high / low prices, and incomplete technical solutions.

[0027] The real-time bidding risk early warning module is linked with the intelligent bidding data analysis module and the adaptive bidding strategy generation module, and sets up a three-level early warning mechanism (general early warning, important early warning, and emergency early warning). For the qualification mismatch, parameter omission (such as unmentioned bidding requirements) found during the analysis process, and risk points found in strategy verification, early warning information is sent according to the risk level. General early warning (such as the omission of non-critical parameters) is prompted through an interface pop-up window, important early warning (such as qualification mismatch) is reminded through in-system message, and emergency early warning (such as the bidding deadline is approaching or the price exceeds the reasonable range) is reminded through both SMS and in-system message.

[0028] The multi-dimensional comparison and analysis module works in conjunction with the three modules mentioned above to conduct multi-dimensional comparisons (such as qualification requirements, scoring standards, and price ranges) between the current bidding project and historical bidding projects and industry benchmark projects, generating a comparison and analysis report to provide a reference for optimizing bidding strategies.

[0029] S3. Data Support Layer Architecture Design: Based on the data support layer requirements in step S1 and the calling requirements of the intelligent core layer in step S2, a distributed data storage architecture is constructed, including a structured database (using MySQL), an unstructured database (using MongoDB), and a real-time cache database (using Redis).

[0030] The structured database stores structured data such as the qualifications of bidding entities (business license, qualification certificate number), historical bidding records (bidding project name, bidding results, price), and key information tag library; the unstructured database stores unstructured data such as bidding documents (PDF, Word format), bid documents (PDF, Word format), and scanned copies of qualification certificates (JPG, PNG format), and adopts a block storage method to improve storage efficiency; the real-time cache database is used to cache temporary data (such as key information in the parsing and intermediate parameters generated by the strategy) and frequently accessed data (such as historical successful bidding cases and industry benchmark data) during the intelligent core layer parsing and calculation process, thereby improving data access speed.

[0031] Simultaneously, a data lifecycle management module is designed to automate the entire process of data collection, cleaning, desensitization, storage, updating, and destruction. The data collection process is linked with the interface adaptation layer, automatically capturing bidding data from external bidding platforms and qualification data from internal enterprise systems. The data cleaning process performs deduplication (removing duplicate bidding information and qualification data), noise reduction (removing invalid characters and erroneous data), missing value imputation (using a fusion algorithm of industry averages and historically similar data; for example, if qualification expiration dates are missing, the average validity period of similar qualifications is used for imputation), and format standardization (unifying files). (Format and data encoding); The data anonymization process irreversibly encrypts the core qualification information of the bidding entity (such as the complete qualification certificate number) and quotation information (using the SHA-256 encryption algorithm) to ensure that the anonymized data cannot be restored, while retaining the statistical analysis value of the data; Data updates adopt a combination of scheduled updates and manual updates. Bidding data is automatically updated once an hour, qualification data and historical bidding data are automatically updated once a month, and special data can be manually triggered for updates; Data destruction is carried out according to the data validity period. Expired bidding data (such as data of unsuccessful projects from 3 years ago) automatically triggers the destruction process, and destruction logs are retained.

[0032] In addition, the real-time cache database adopts a cache eviction strategy, which constructs eviction rules based on data call frequency (number of calls in the past 7 days), data timeliness (such as the validity period of bidding information), and data importance (core data, temporary data). It automatically evicts temporary data that is called less than 5 times / 7 days, exceeds the validity period, or is not core. The preset threshold is linked with the data analysis module of the intelligent core layer and is adaptively adjusted according to historical call data. For example, during the bidding peak period, the call frequency threshold is reduced to ensure the cache stability of high-frequency data.

[0033] S4. Collaborative Design of Interface Adaptation Layer and Interaction Layer: Based on the requirements of the interface adaptation layer and user interaction layer in step S1, and combined with the module functions of the intelligent core layer and data support layer, design the multi-protocol adaptive interface module of the interface adaptation layer and the multi-terminal adaptation interface of the user interaction layer.

[0034] The multi-protocol adaptive interface module supports multiple protocols such as HTTP, HTTPS, and WebService. It integrates an interface detection module and a protocol parsing module. The interface detection module detects the communication protocols of external systems (such as external bidding platforms and internal ERP systems) in real time, automatically matching the corresponding interface protocol and adaptively adjusting data interaction parameters (such as data transmission rate and format). If an unknown protocol is detected, the protocol parsing module is automatically triggered. Based on the protocol feature library (which stores characteristic parameters of common communication protocols) in the data support layer, it parses the protocol and generates a temporary adaptive interface to ensure the continuity of data interaction. The protocol feature library is linked with the data support layer; it is automatically updated each time a new interface protocol is added. This module also works in conjunction with the data acquisition module in the data support layer to automatically capture bidding data (retrieving bidding announcements and bidding documents from external bidding platforms) and automatically upload bid data (uploading bid documents and quotation information to external bidding platforms). Simultaneously, it enables data interaction with internal enterprise systems, capturing qualification data, cost data, etc.

[0035] The user interaction layer features a responsive interface that seamlessly switches between PC and mobile devices (phones, tablets). The interface layout adapts to different device sizes and integrates functions such as bidding data visualization, strategy editing, risk monitoring, and command issuance. It also works in conjunction with other modules in the intelligent core layer: the visualization module displays bidding progress, risk levels, and comparative analysis results using charts (line graphs, bar charts, pie charts); the strategy editing module allows users to manually modify intelligently generated bidding strategies and save modification records; the risk monitoring module displays all warning information and allows viewing warning details and handling suggestions; and the command issuance module supports users issuing commands for data collection, file uploads, and strategy generation.

[0036] In addition, the multi-terminal adapted interface also integrates a bidding progress visualization tracking module. This module works in conjunction with the intelligent core layer and data support layer to capture the progress data of each stage of the bidding process (document preparation, qualification review, quotation confirmation, and submission and upload) in real time. It displays the completion status and remaining time of each stage in the form of a timeline, and sets progress node warnings. When a node is not completed within the time limit (such as document preparation timeout), the real-time bidding risk warning module is triggered to send a reminder message to the user to ensure that the bidding process proceeds on time.

[0037] S5. Embedded design of security protection layer: Based on the security protection layer requirements in step S1, and combined with the functional characteristics of each layer architecture, the security protection module is embedded and integrated into the user interaction layer, intelligent core layer, data support layer, and interface adaptation layer to achieve comprehensive security protection.

[0038] The permission hierarchical control module is linked with the user interaction layer, and different operation permissions are assigned based on the bidding role (administrator, bidding specialist, reviewer). Administrators have full permissions (architecture configuration, permission management, data viewing), bidding specialists have operation permissions (strategy editing, document preparation, data uploading), and reviewers have review permissions (strategy review, document review). The RBAC permission management model is adopted to achieve fine-grained control of permissions and prevent unauthorized operations. The data encryption transmission module works in conjunction with the interface adaptation layer and the data support layer. The data transmission process uses the SSL encryption protocol to achieve end-to-end encryption and prevent theft or tampering during data transmission. The data storage process uses the AES-256 encryption algorithm to encrypt and store core data (quotations, qualification information). The key is managed by the administrator and changed regularly. The operation log auditing module works in conjunction with other modules to record all operations, including the operator, operation time, operation content, and operation results. The log data is stored in a structured database and retained for 6 months. It supports retrieval by operator, operation time, and operation type for traceability and verification. In the event of data leakage or operational errors, the responsible party can be quickly identified. The abnormal behavior interception module works in conjunction with the intelligent core layer. Based on the historical operation data of the intelligent core layer, it constructs a normal operation behavior characteristic model (such as operation time, operation frequency, and data access range). It collects current operation behavior data in real time and compares it with the normal behavior characteristic model. The preset deviation threshold is 30%. When the deviation exceeds 30%, it is judged as abnormal behavior (such as malicious multiple login attempts or batch downloading of core data). An interception command is immediately triggered to prohibit the operation. A warning message is sent to the administrator through the real-time bidding risk warning module. The administrator can manually lift the interception or conduct further verification. The preset threshold can be adaptively adjusted according to the actual application scenario. For example, during peak bidding periods, the deviation threshold can be appropriately increased.

[0039] S6. Architecture Linkage Debugging and Iterative Optimization: Integrate the architecture modules designed in steps S2-S5, and use Spring Cloud microservice architecture to achieve integrated deployment of each module, thus building a complete intelligent bidding assistance system architecture.

[0040] By simulating the entire bidding process (selecting three different types of bidding projects, covering engineering, services, and goods), the performance of the linkage between each layer of the architecture was debugged, and the stability of interface adaptation was verified (testing the success rate of integration with five different external bidding platforms, requiring a success rate of ≥99%), the timeliness of data calls (testing the response time of high-frequency data calls, requiring a response time of ≤1s), the accuracy of intelligent strategies (comparing the bidding matching degree between intelligent generation strategies and manually optimized strategies, requiring a matching degree of ≥90%), and the reliability of security protection (simulating malicious attacks and unauthorized operations, testing the interception success rate, requiring an interception success rate of ≥99.5%). Problems encountered during the debugging process were recorded, such as interface integration lag, large deviations in strategy generation, and untimely warnings.

[0041] Based on the debugging results and combined with the requirements of step S1 - the module association matrix, the functional parameters of each module were iteratively optimized: to address the interface connection bottleneck issue, the parameters of the multi-protocol adaptive interface module were optimized to improve the data transmission rate; to address the issue of large deviations in strategy generation, the amount of historical bidding case data was increased and the machine learning model parameters were optimized; to address the issue of untimely warnings, the warning triggering mechanism of the real-time warning module for bidding risks was adjusted to shorten the warning response time.

[0042] Meanwhile, a closed-loop process of "design-debugging-optimization-iteration" is established. After the system is officially launched, based on the data call time, strategy generation accuracy, risk warning accuracy, and user operation feedback data during the operation process, multi-dimensional statistical analysis is carried out through the data analysis module of the intelligent core layer. An optimization report for each layer module is generated once a month. The report includes module functional defects, parameter adjustment suggestions, and performance improvement plans. Based on the optimization report, targeted adjustments are made to each layer module to continuously improve the adaptability and stability of the architecture.

[0043] The intelligent bidding assistance system in this embodiment features a five-layer architecture, intelligent module design, comprehensive security protection, and closed-loop iterative optimization. This architecture provides intelligent support for the entire bidding process. Tests have shown that it can improve the efficiency of parsing bidding documents by more than 80%, shorten the time for generating bidding strategies to less than 30 minutes, achieve a bidding risk warning accuracy rate of ≥95%, and increase the bidding success rate by more than 30%. This significantly reduces labor costs and bidding risks, and is suitable for the bidding needs of enterprises of different types and sizes. It has good practicality and promotional value.

[0044] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for designing the overall architecture of an intelligent bidding assistance system, characterized in that, Includes the following steps: S1. Layered Analysis of Architectural Requirements: Based on the entire bidding process, the basic architectural requirements are broken down into five layers: user interaction layer, intelligent core layer, data support layer, interface adaptation layer, and security protection layer. Specifically, the user interaction layer corresponds to the bidding entity's operational scenarios; the intelligent core layer corresponds to bidding data analysis, strategy generation, and risk warning scenarios; the data support layer corresponds to bidding data storage, retrieval, and update scenarios; the interface adaptation layer corresponds to data interaction scenarios with external bidding platforms and internal enterprise management systems; and the security protection layer corresponds to bidding data confidentiality and operational permission control scenarios. Furthermore, a requirement priority ranking algorithm is used to filter the core requirements of each layer, establish a mapping relationship between requirements and architectural modules, and generate a requirement-module association matrix to provide a basis for subsequent architecture design. S2. Intelligent Core Layer Module Design: Based on the core requirements of the intelligent core layer obtained in step S1, an intelligent core layer is constructed, consisting of a bid data intelligent parsing module, a bid strategy adaptive generation module, a bid risk real-time early warning module, and a multi-dimensional comparative analysis module. The bid data intelligent parsing module is equipped with an OCR recognition and NLP semantic analysis linkage algorithm to parse key parameters, qualification requirements, and scoring criteria in bidding announcements and bidding documents. The bid strategy adaptive generation module is linked with the bid data intelligent parsing module, generating personalized bid strategies adapted to the current bidding project based on the parsed key information, combined with historical bidding data and industry benchmark data. The bid risk real-time early warning module is linked with both the bid data intelligent parsing module and the bid strategy adaptive generation module, providing tiered early warnings for qualification mismatches, parameter omissions, and unreasonable strategies discovered during the parsing process. S3. Data Support Layer Architecture Design: Based on the data support layer requirements in step S1 and the data call requirements of each module in the intelligent core layer in step S2, a distributed data storage architecture is constructed. This distributed data storage architecture includes a structured database, an unstructured database, and a real-time cache database. The structured database stores structured data such as the qualifications of bidding entities and historical bidding records. The unstructured database stores unstructured data such as bidding documents, tender documents, and scanned copies of qualification certificates. The real-time cache database caches temporary data and frequently called data during the parsing and calculation process of the intelligent core layer. Simultaneously, a data lifecycle management module is designed to automate the entire process of data collection, cleaning, desensitization, storage, updating, and destruction. The data collection stage is linked with the interface adaptation layer, and the data cleaning and desensitization stages provide high-quality data support for the intelligent core layer. S4. Collaborative Design of Interface Adaptation Layer and Interaction Layer: Based on the requirements of the interface adaptation layer and user interaction layer in step S1, and combined with the module functions of the intelligent core layer and data support layer in steps S2 and S3, a multi-protocol adaptive interface module for the interface adaptation layer is designed. This multi-protocol adaptive interface module supports multiple protocols such as HTTP, HTTPS, and WebService, enabling seamless data interaction with external bidding platforms and internal enterprise management systems. It also works in conjunction with the data acquisition module of the data support layer to automatically capture bidding data and automatically upload bid data. Simultaneously, a multi-terminal adaptive interface for the user interaction layer is designed. This interface works in conjunction with the modules of the intelligent core layer to realize the visualization of bid data, strategy editing, risk viewing, and operation command issuance, supporting seamless switching between PC, mobile, and tablet terminals. S5. Embedded Design of Security Protection Layer: Based on the security protection layer requirements in step S1, and combined with the functional characteristics of each layer architecture in steps S2-S4, the security protection module is embedded and integrated into each layer architecture. The security protection module includes a permission hierarchical control module, a data encryption transmission module, an operation log auditing module, and an abnormal behavior interception module. Among them, the permission hierarchical control module is linked with the user interaction layer to assign different operation permissions based on the bidding role; the data encryption transmission module is linked with the interface adaptation layer and the data support layer to realize end-to-end encryption during data transmission and storage; the operation log auditing module is linked with each layer module to record all operation behaviors for traceability and verification; the abnormal behavior interception module is linked with the intelligent core layer to intercept malicious operations and abnormal data access based on behavioral feature analysis. S6. Architecture Linkage Debugging and Iterative Optimization: Integrate the architecture modules designed in steps S2-S5 to construct a complete intelligent bidding assistance system architecture; debug the linkage performance between the architecture layers by simulating the entire bidding process, verifying the stability of interface adaptation, the timeliness of data calls, the accuracy of intelligent strategies, and the reliability of security protection; based on the debugging results, combined with the requirement-module association matrix in step S1, iteratively optimize the functional parameters of each module to form a feasible and highly adaptable intelligent bidding assistance system architecture. The iterative optimization process is linked with the data analysis module of the intelligent core layer, automatically generating optimization suggestions based on debugging data and historical operating data.

2. The overall architecture design method of an intelligent bidding assistance system according to claim 1, characterized in that, The requirement priority ranking algorithm mentioned in step S1 is as follows: a three-dimensional evaluation system is constructed based on the impact of bidding business, the urgency of user needs and the difficulty of technical implementation. The comprehensive priority score of each requirement is calculated by the analytic hierarchy process, and requirements with a comprehensive score ≥ 80 are selected as core requirements. In the three-dimensional evaluation system, the weight of business impact is 0.4, the weight of requirement urgency is 0.3, and the weight of technical implementation difficulty is 0.

3.

3. The overall architecture design method of an intelligent bidding assistance system according to claim 1, characterized in that, The OCR recognition and NLP semantic analysis linkage algorithm described in step S2 specifically includes: extracting text information from text, tables, and images in the bidding documents through the OCR recognition module to generate standardized text; the NLP semantic analysis module performing word segmentation, part-of-speech tagging, and entity recognition on the standardized text to extract key information such as the bidding project name, bidding unit, qualification requirements, scoring rules, and bid deadline, and establishing a key information tag library. The key information tag library is linked with the structured database of the data support layer to realize rapid retrieval and access of key information.

4. The overall architecture design method of an intelligent bidding assistance system according to claim 1, characterized in that, The personalized bidding strategy described in step S2 is generated as follows: the adaptive bidding strategy generation module calls historical successful bidding case data and industry benchmark data in the data support layer, compares them with the key information of the current bidding project, constructs a strategy generation model through machine learning algorithms, and outputs qualification matching suggestions, price range suggestions, technical solution optimization suggestions, and key suggestions for bid document preparation. The parameters of the strategy generation model are continuously trained and optimized through historical bidding data, and are linked with the real-time bidding risk early warning module to perform risk verification on the generated bidding strategy.

5. The overall architecture design method of an intelligent bidding assistance system according to claim 1, characterized in that, The data lifecycle management module described in step S3 specifically includes the following data cleaning process: The collected bidding-related data undergoes deduplication, noise reduction, missing value filling, and format standardization. Missing value filling employs a fusion filling algorithm based on industry averages and historically similar data. The data anonymization process involves encrypting and anonymizing the core qualification information and pricing information of the bidding entity using an irreversible encryption algorithm to ensure that the anonymized data cannot be restored while retaining its statistical analysis value.

6. The overall architecture design method of an intelligent bidding assistance system according to claim 1, characterized in that, The multi-protocol adaptive interface module mentioned in step S4 has the function of automatic protocol identification and switching. Specifically, it detects the communication protocol of the external interface system through the interface detection module, automatically matches the corresponding interface protocol, and completes the adaptive adjustment of data interaction parameters. If an unknown protocol is detected, the protocol parsing module is automatically triggered to parse based on the protocol feature library and generate a temporary adaptive interface to ensure the continuity of data interaction. The protocol feature library is linked with the data support layer to achieve real-time updates.

7. The overall architecture design method of an intelligent bidding assistance system according to claim 1, characterized in that, The abnormal behavior interception module described in step S5 has the following behavior feature analysis process: Based on the historical operation data of the intelligent core layer, a normal operation behavior feature model is constructed, the current operation behavior data is collected in real time and compared with the normal behavior feature model. When the comparison deviation exceeds a preset threshold, it is determined to be abnormal behavior, an interception command is immediately triggered, and a warning message is sent to the administrator through the real-time bidding risk warning module. The preset threshold can be adaptively adjusted according to the actual application scenario.

8. The overall architecture design method of an intelligent bidding assistance system according to claim 1, characterized in that, The iterative optimization described in step S6 specifically includes: based on the data call time, strategy generation accuracy, risk warning accuracy, and user operation feedback data during system operation, multi-dimensional statistical analysis is performed through the data analysis module of the intelligent core layer to generate optimization reports for each layer of modules. The optimization reports include module functional defects, parameter adjustment suggestions, and performance improvement plans. Targeted adjustments are made to each layer of modules according to the optimization reports to form a closed-loop process of "design-debugging-optimization-iteration".

9. The overall architecture design method of an intelligent bidding assistance system according to claim 1, characterized in that, The multi-terminal adaptation interface mentioned in step S4 also integrates a bidding progress visualization tracking module. The tracking module is linked with the intelligent core layer and the data support layer to capture the progress data of each stage of bidding (document preparation, qualification review, quotation confirmation, submission and upload) in real time and display it in the form of a timeline. At the same time, progress node warnings are set. When a node is not completed within the time limit, the bidding risk real-time warning module is triggered to send a reminder message.

10. The overall architecture design method of an intelligent bidding assistance system according to claim 1, characterized in that, The real-time cache database mentioned in step S3 adopts a cache eviction strategy. The cache eviction strategy is constructed based on data call frequency, data timeliness and data importance. It automatically evicts temporary data that has a call frequency lower than a preset threshold, has exceeded its validity period or is not core, so as to ensure the storage efficiency of the cache database and the timeliness of data call. The preset threshold is linked with the data analysis module of the intelligent core layer and is adaptively adjusted according to historical call data.

Images