Electronic file assisted intelligent bidding process management method and system

Abstract

The application relates to the technical field of bidding process management, and provides an intelligent bidding process management method and system assisted by electronic archives, which comprises the following steps: obtaining running state information of a storage medium of the electronic archives and integrity verification records of the electronic archives according to storage location information of the electronic archives; evaluating the determination degree of a processing procedure for extracting key information from the electronic archives; comparing the extracted key information with other related information of a supplier and industry reference information to obtain a fitting degree; calculating a credibility measure value according to the running state information of the storage medium, the integrity verification records, the determination degree of the processing procedure and the fitting degree; generating a verification prompt when the credibility measure value is lower than a preset judgment standard; providing an interactive interface to receive a result of manual review of original electronic archives pointed by the verification prompt, and updating a final credibility state of the key information according to the review result. The application can improve the accuracy of bidding decision.

Description

Technical Field

[0001] This application relates to the field of bidding process management technology, and more specifically, to an intelligent bidding process management method and system assisted by electronic records. Background Technology

[0002] Enterprises commonly adopt intelligent bidding process management systems assisted by electronic records to improve efficiency and decision-making quality. These systems need to process massive amounts of electronic records, including supplier qualifications, performance records, and technical solutions, and continuously write them to core data storage devices. However, under prolonged high-load operation, some hard drives in the storage devices may enter a sub-optimal state. This insidious degradation, especially in specific scenarios involving high-concurrency small file writes, may lead to temporary inconsistencies between the physical storage of electronic records and the logical state of the system.

[0003] This inconsistency can lead to the system reading data that is not fully synchronized or contains logical deviations when generating checksums for electronic documents, resulting in erroneous checksums. These erroneous checksums are like altered digital fingerprints, causing deviations in subsequent intelligent bidding processing logic when parsing the document content. For example, the intelligent system may fail to accurately extract key technical parameters or correctly understand text content, leading to subtle but critical errors in the construction of supplier profiles or risk assessments. The accumulation of such deviations may cause companies to miss out on more innovative or cost-effective potential partners. Summary of the Invention

[0004] This application provides an intelligent bidding process management method and system assisted by electronic records to solve at least one of the above-mentioned technical problems.

[0005] To achieve the above objectives, this application adopts the following technical solution:

[0006] Firstly, this application discloses an intelligent bidding process management method assisted by electronic records, including:

[0007] Obtain the storage location information of electronic records;

[0008] Based on the storage location information, obtain the operational status information of the storage medium of the electronic archives and the integrity verification records of the electronic archives;

[0009] Assess the degree of certainty in the process of extracting key information from electronic records;

[0010] The extracted key information is compared with other relevant information from the supplier and industry reference information to obtain the degree of consistency between the extracted key information and other relevant information and industry reference information.

[0011] Based on the storage medium's operational status information, integrity verification records, the degree of certainty and consistency of the processing procedure, calculate the credibility metric of the extracted key information;

[0012] When the credibility metric value is lower than the preset judgment standard, a verification prompt is generated, which includes key information, the credibility metric value, the reason why the credibility metric value is lower than the preset judgment standard, and the suggested scope of verification.

[0013] By incorporating verification prompts into the bid evaluation report, the underlying data uncertainties that machines cannot resolve independently can be transformed into human understanding and operational verification.

[0014] Provide an interactive interface to receive the results of manual review of the original electronic files indicated by the verification prompts, and update the final credibility status of key information based on the review results;

[0015] The final credibility status of key information is displayed through an interactive interface to serve as auxiliary information for human decision-making.

[0016] Secondly, this application also discloses an intelligent bidding process management system assisted by electronic records, the system comprising:

[0017] The storage location information acquisition module is used to acquire the storage location information of electronic archives;

[0018] The operation status information acquisition module is used to acquire the operation status information of the storage medium of the electronic archive and the integrity verification record of the electronic archive based on the storage location information;

[0019] The certainty assessment module is used to assess the degree of certainty in the process of extracting key information from electronic archives;

[0020] The matching degree comparison module is used to compare the extracted key information with other relevant information of the supplier and industry reference information to obtain the degree of matching between the extracted key information and other relevant information and industry reference information.

[0021] The trust metric calculation module is used to calculate the trust metric of the extracted key information based on the storage medium's operating status information, integrity verification records, the degree of certainty of the processing process, and the degree of consistency.

[0022] The verification prompt generation module is used to generate a verification prompt containing key information, the credibility value, the reason why the credibility value is lower than the preset judgment standard, and the suggested verification scope when the credibility metric value is lower than the preset judgment standard.

[0023] The bid evaluation report module is used to integrate verification prompts into the bid evaluation report, thereby transforming the underlying data uncertainties that machines cannot resolve independently into human understanding and operational verification.

[0024] The review result update module provides an interactive interface for receiving the results of manual review of the original electronic files indicated by the review prompts, and updates the final credibility status of key information based on the review results.

[0025] The display module is used to show the final credibility status of key information through an interactive interface, serving as auxiliary information for human decision-making.

[0026] Compared with the prior art, this application has at least the following beneficial effects:

[0027] This application significantly improves the accuracy, reliability, and traceability of bidding and evaluation results, providing a solid technical guarantee for enterprises to make more informed decisions in bidding activities. Attached Figure Description

[0028] Figure 1 A flowchart illustrating an intelligent bidding process management method assisted by electronic records, provided in this application;

[0029] Figure 2 This is a schematic diagram of the structure of an intelligent bidding process management system with electronic file assistance provided in this application. Detailed Implementation

[0030] The technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0031] like Figure 1 As shown in the figure, this application proposes an intelligent bidding process management method assisted by electronic archives, including:

[0032] Obtain the storage location information of electronic records;

[0033] Based on the storage location information, obtain the operational status information of the storage medium of the electronic archives and the integrity verification records of the electronic archives;

[0034] Assess the degree of certainty in the process of extracting key information from electronic records;

[0035] The extracted key information is compared with other relevant information from the supplier and industry reference information to obtain the degree of consistency between the extracted key information and other relevant information and industry reference information.

[0036] Based on the storage medium's operational status information, integrity verification records, the degree of certainty and consistency of the processing procedure, calculate the credibility metric of the extracted key information;

[0037] When the credibility metric value is lower than the preset judgment standard, a verification prompt is generated, which includes key information, the credibility metric value, the reason why the credibility metric value is lower than the preset judgment standard, and the suggested scope of verification.

[0038] By incorporating verification prompts into the bid evaluation report, the underlying data uncertainties that machines cannot resolve independently can be transformed into human understanding and operational verification.

[0039] Provide an interactive interface to receive the results of manual review of the original electronic files indicated by the verification prompts, and update the final credibility status of key information based on the review results;

[0040] The final credibility status of key information is displayed through an interactive interface to serve as auxiliary information for human decision-making.

[0041] Electronic archives refer to various digital documents generated during bidding activities, such as supplier qualification certificates, technical proposals, commercial quotations, contract texts, and performance records. These archives are usually stored in computer systems in file form and are core data for bidding process management.

[0042] Storage location information refers to the specific storage path or identifier of electronic records in a storage system, such as file path, database record ID, object key in a distributed storage system, etc. This information can be used to locate the physical or logical storage location of electronic records.

[0043] Storage media operating status information refers to the real-time operating status data of hardware devices (such as hard drives, solid-state drives, storage arrays, etc.) that store electronic records, including but not limited to read / write speed, error rate, temperature, health indicators (such as SMART information), etc. This information reflects the health status and performance of the storage media.

[0044] Integrity verification records are records used to verify the contents of electronic documents to ensure that they have not been tampered with or damaged. They are usually generated and stored using hash values ​​(such as MD5, SHA-256) or digital signatures.

[0045] Key information refers to data points extracted from electronic files that have a significant impact on bidding decisions, such as supplier name, registered capital, technical parameters, project experience, and bid amount.

[0046] The credibility metric is a comprehensive numerical value used to quantify the extent to which extracted key information is reliable, accurate, and trustworthy. This value typically ranges from 0 to 1, with higher values ​​indicating greater credibility.

[0047] A verification prompt is a system-generated report or message that reminds humans to review certain key information with low credibility. It includes the background information and guidance required for the review.

[0048] The bid evaluation report is a formal document used in the bidding process to record the bid evaluation process, evaluation results, and recommendations for winning candidates.

[0049] An interactive interface refers to a graphical user interface (GUI) provided by a system for users to operate and display information. Through this interface, users can input commands, view data, and perform reviews, etc.

[0050] The following is a further detailed description of the electronically-assisted intelligent bidding process management method provided in this application:

[0051] The system needs to obtain the storage location information of electronic archives. This can be achieved in several ways. For example, the system can maintain a metadata database for electronic archives, containing a unique identifier for each archive and its corresponding storage path. When a particular electronic archive needs to be processed, the system can query this database to obtain its storage location information. Another approach is that when an electronic archive is uploaded or created, the system records its storage location in real time and associates it with the archive content. For example, in a file system, the full path of the file can be directly obtained; in a database, the table name and record ID storing the archive can be retrieved.

[0052] Based on the acquired storage location information, the system obtains the operational status information of the electronic archive's storage medium and its integrity verification records. For the storage medium's operational status information, the system can obtain it in real time through integrated storage device monitoring interfaces (such as SNMP or API). For example, it can periodically query the storage array controller to obtain SMART (Self-Monitoring, Analysis and Reporting Technology) data for the hard drives, including error rate, remapped sector count, and temperature. This data reflects the health status of the storage medium. For integrity verification records, the system typically calculates and stores the hash value (such as MD5 or SHA-256) of the electronic archive when it is generated or stored. When needed, the system recalculates the hash value of the current electronic archive and compares it with the stored verification record to determine if the archive has been tampered with or corrupted. For example, when a PDF document is uploaded, the system immediately calculates and stores its SHA-256 hash value. Subsequent processing of the document will recalculate the hash value and compare it with the previously stored value.

[0053] The system assesses the degree of certainty in the process of extracting key information from electronic documents. This assessment aims to measure the reliability of information extraction. For example, if key information is read directly from a structured table, its degree of certainty is likely to be high. However, if key information is inferred from unstructured text using Natural Language Processing (NLP) techniques, its degree of certainty is likely to be low. One implementation approach is for the system to pre-define a set of rules to assign an initial degree of certainty based on the source and method of information extraction. For example, data extracted from PDF form fields has a degree of certainty of 0.9; data extracted from free text paragraphs through keyword matching has a degree of certainty of 0.7; and data inferred through complex semantic analysis has a degree of certainty of 0.5.

[0054] The system compares the extracted key information with other relevant information about the supplier and industry reference information to determine the degree of similarity between the extracted key information and other relevant information and industry reference information. Other relevant information about the supplier may include its historical performance records, financial statements, and corporate credit reports. Industry reference information may include industry averages, market prices, and technical standards. The comparison can be conducted in various ways. For example, for numerical data, numerical comparisons can be performed directly to calculate the deviation rate. For textual data, semantic similarity analysis can be performed. For example, if the extracted key information is that supplier A's registered capital is 10 million yuan, the system will compare it with the registered capital recorded in supplier A's business registration information and with the average registered capital of companies in the same industry to determine the degree of similarity.

[0055] The system calculates a credibility metric for the extracted key information based on the storage medium's operational status, integrity verification records, the certainty of the processing procedure, and the degree of consistency. This is a comprehensive evaluation process. For example, a weight can be assigned to each evaluation dimension, and then the evaluation results of each dimension can be weighted and summed. For instance, if the storage medium is in good operating condition, integrity verification passes, the certainty of information extraction is high, and the consistency with external information is high, the credibility metric will be high. Conversely, if a problem occurs in a certain dimension, such as the storage medium having a large number of bad blocks, or integrity verification failing, the credibility metric will be significantly reduced.

[0056] When the calculated trust metric value is lower than the preset judgment standard, the system will generate a verification prompt containing identification information of key information, the trust metric value, the reason why the trust metric value is lower than the preset judgment standard, and a suggested scope of verification. For example, the preset judgment standard could be 0.6. If the trust metric value of a certain key piece of information is 0.5, the system will generate a verification prompt. The verification prompt will clearly indicate which key piece of information is being verified (e.g., the description of memory capacity in vendor B's technical solution), its trust metric value, why it is low (e.g., potential failure risk of the storage medium and low certainty of information extraction), and the suggested scope of verification (e.g., reviewing section 3.2 of the original technical solution document and checking the storage device logs).

[0057] The system integrates verification prompts into the bid evaluation report, transforming underlying data uncertainties that machines cannot resolve independently into matters requiring human understanding and verification. This means that verification prompts are no longer standalone warnings but are presented to the bid evaluation committee as part of the report. For example, in the supplier technical evaluation section, if a supplier's technical parameters have credibility issues, the system will include a link or summary of a verification prompt next to that parameter, reminding evaluators to pay attention and conduct manual verification. In this way, evaluators can clearly see which information requires special attention and human intervention when reading the report.

[0058] The system also provides an interactive interface for receiving the results of manual review of the original electronic files indicated by the verification prompts, and updating the final credibility status of key information based on the review results. This interface allows evaluation personnel to directly view the original electronic files and correct or confirm key information according to their own judgment. For example, evaluation personnel can open the original technical solution document on the interface, manually check the description of memory capacity, and if they find an error in the system's extraction, they can directly correct the value on the interface. After correction, the system will reassess and update the final credibility status of the key information based on the manual review results.

[0059] The system will display the final credibility status of key information through the interactive interface as supplementary information for human decision-making. This means that key information, after human review and updates, and its final credibility status will be clearly presented on the interactive interface. For example, in the supplier comprehensive evaluation interface, the final credibility status of each key indicator will be displayed next to it (e.g., manually confirmed, risk exists but has been manually corrected), providing evaluation personnel with a more comprehensive and reliable basis for decision-making.

[0060] This application conducts a multi-dimensional analysis of the reliability of key information from the perspectives of accuracy of information extraction and external consistency of information content. These multi-dimensional evaluation results are comprehensively used to calculate the credibility metric of the extracted key information. This comprehensive calculation mechanism makes the judgment of the reliability of key information no longer single-dimensional, but comprehensive and detailed. When the credibility metric is lower than the preset judgment standard, the system can intelligently generate a verification prompt. This prompt not only includes the identification information and credibility metric of the key information, but more importantly, it clearly points out the reason for the low credibility metric and the suggested scope of verification. This effectively transforms the underlying data uncertainties that machines cannot resolve independently into specific guidance that humans can understand and operate for review.

[0061] By integrating verification prompts into the bid evaluation report, this application ensures that the bid evaluation committee can promptly and intuitively identify potential data risks during report review, thereby avoiding decisions based on uncertain data. Furthermore, providing an interactive interface to receive human review results and update the final credibility status of key information forms a closed-loop feedback mechanism, enabling human expertise to effectively correct and refine the machine's initial assessment. Finally, displaying the final credibility status of key information through the interactive interface provides humans with decision-making support information that has undergone multiple verifications and corrections, significantly improving the accuracy and reliability of bidding decisions.

[0062] This application assesses the reliability of electronic archives directly from the data source by introducing the acquisition of storage medium operational status information and electronic archive integrity verification records. Furthermore, the assessment of the certainty of the information extraction process and the comparison of consistency with multi-source information further enhance the dimensions for judging the reliability of key information. More importantly, this application integrates these assessment results into a credibility metric and generates detailed verification prompts when credibility is insufficient, transforming the underlying data uncertainty that machines cannot resolve independently into a concrete task that can be understood and operated by humans for review. This collaborative working mode of machine initial assessment + human review not only improves the accuracy of decision-making but also makes the risk points in the decision-making process transparent and traceable. Receiving the human review results through an interactive interface and updating the final credibility status of key information forms an effective feedback loop, ensuring the reliability of the final decision-making support information. Therefore, this application demonstrates significant technological progress and practical value in addressing the risks of intelligent bidding decision-making caused by underlying data uncertainty.

[0063] In some embodiments, the above-described electronically assisted intelligent bidding process management method further includes:

[0064] At the controller level of the storage array, micro-behavioral data of every input and output operation is captured. This micro-behavioral data includes instruction response time, instantaneous data transfer rate, frequency of internal error correction mechanism triggering, and write confirmation latency fluctuations.

[0065] For each storage cell in the storage array, establish a normal behavioral fingerprint describing the distribution of its microscopic input and output behavioral characteristics under different load conditions;

[0066] When the captured micro-behavioral data of input and output operations deviates statistically from the normal behavioral fingerprint of the storage unit, the data block involved in the operation is marked as transiently high risk;

[0067] Adjust the source reliability factor of electronic records based on the instantaneous high-risk markers;

[0068] The adjusted source reliability factor is applied to the calculation of the credibility metric value of the extracted key information.

[0069] Specifically, at the controller level of the storage array, by deploying dedicated monitoring agents or utilizing the storage system's built-in performance monitoring interfaces, micro-behavioral data of every input and output operation can be captured in real time. This micro-behavioral data is a key indicator for measuring the health of the storage system and the quality of data processing. For example, instruction response time reflects the storage system's processing speed for requests; instantaneous data transfer rate reveals the smoothness of data flow; the trigger frequency of internal error correction mechanisms directly indicates the occurrence of data errors within the storage medium; and fluctuations in write confirmation latency may suggest instability in storage write operations. This real-time acquisition of data aims to provide more detailed and immediate insights into storage behavior than traditional macro-level operational status information.

[0070] Establishing a normal behavior fingerprint for each storage cell in a storage array involves building a baseline model through long-term, multi-load condition data collection and statistical analysis. This model describes the distribution patterns of the micro-input and output behavior characteristics of the storage cell under normal operating conditions. For example, machine learning algorithms can be used to train on historical data to learn the typical ranges and fluctuation patterns of instruction response time, data transfer rate, etc., under different load modes. The purpose is to provide a reliable reference standard for subsequent anomaly detection.

[0071] When the captured micro-behavioral data of input and output operations deviates statistically from the normal behavioral fingerprint of the storage unit—for example, a sudden and significant increase in instruction response time or an abnormally high frequency of internal error correction mechanism triggering—it indicates that the operation may be abnormal. In this case, the data block involved in the operation will be marked as transiently high-risk. This marking is real-time and designed to quickly identify potential data integrity or reliability issues.

[0072] Based on transient high-risk markers, the source reliability factor of electronic records can be dynamically adjusted. For example, when a data block stored in an electronic record is marked as transiently high-risk, its corresponding source reliability factor will be reduced. This source reliability factor is a quantitative indicator used to measure the reliability of the original storage environment and process of electronic records. Its purpose is to transform micro-anomalies in the underlying storage system into inputs for assessing the credibility of the upper-level electronic records.

[0073] The adjusted source reliability factor will be applied to the calculation of the credibility metric for the extracted key information. This means that when calculating the credibility of key information, in addition to considering factors such as the operational status information of the storage medium, integrity verification records, the certainty of the processing process, and the degree of consistency, a more refined assessment of the source reliability of electronic archives will also be incorporated, so that the final credibility metric can more comprehensively and accurately reflect the true reliability of the key information.

[0074] This application captures micro-behavioral data at the controller level of the storage array and compares it with pre-established normal behavioral fingerprints, enabling real-time and precise identification of potential transient anomalies at the underlying storage system level. This deep insight into micro-behavior allows the system to promptly detect potential data risks that are difficult to detect with traditional macro-monitoring. When a statistically significant deviation is detected, the relevant data block is marked as transiently high-risk, directly triggering an adjustment to the reliability factor of the electronic record source. By incorporating this dynamically adjusted source reliability factor into the calculation of the credibility metric for critical information, this application effectively links the underlying health condition of the storage medium with the reliability assessment of the upper-level critical information. This mechanism ensures that even transient, localized storage anomalies are promptly reflected in the credibility assessment of critical information, thereby avoiding misjudgments caused by problems with the underlying data source.

[0075] In some embodiments, the steps described above for assessing the degree of certainty in the process of extracting key information from electronic archives include:

[0076] Identify technical terms or abbreviations in key information extracted from electronic archives, and analyze the logical roles and semantic relationships of these terms or abbreviations in the current sentences, paragraphs, and documents.

[0077] Based on all the bidding documents in the current batch, construct an association network consisting of high-frequency professional terms or abbreviations and their contextual roles, as a domain consensus semantic framework.

[0078] The extracted key information is compared with the domain consensus semantic framework in a deep semantic context consistency comparison. The logical roles and semantic relationships of professional terms or abbreviations in the key information are evaluated to determine whether they are consistent with the domain consensus semantic framework, whether the related information is logically reasonable, and whether the meaning expressed deviates from the general understanding of the domain.

[0079] Based on the results of deep semantic context consistency comparison, the degree of certainty in the information extraction process is adjusted, specifically as follows:

[0080] When there is a risk of inconsistency in the deep semantic context, the certainty of information extraction is reduced, and risk warnings containing specific contradictions or deviations are generated.

[0081] Specifically, identifying technical terms or abbreviations in electronic documents involves using natural language processing techniques, such as lexical analysis and named entity recognition, to extract relevant technical terms or their abbreviations from the text content of electronic documents. Simultaneously, contextual analysis is performed on these identified terms or abbreviations to determine their grammatical functions (such as subject, predicate, object, and modifier) ​​within specific sentences, paragraphs, or even the entire document, as well as their semantic relationships with other words, such as causal, parallel, and subordinate relationships. The aim is to provide a foundation for subsequent deep semantic contextual consistency comparison.

[0082] Based on all bidding documents in the current batch, a relational network consisting of high-frequency technical terms or abbreviations and their contextual roles is constructed as a domain consensus semantic framework. This can be understood as the system learning from and analyzing a large number of bidding documents to extract frequently occurring technical terms or abbreviations within the domain, and recording the logical roles and semantic relationships these terms or abbreviations typically play in different contexts. For example, a certain technical term may appear as the object of a specific action in most cases, or it may always have a specific modifying relationship with another technical term. This relational network constitutes the common sense or standard semantic model of the bidding domain, used to measure the normativity of subsequent information extraction.

[0083] Deep semantic contextual consistency comparison of extracted key information with a domain consensus semantic framework involves comparing key information extracted from a single electronic file, including technical terms or abbreviations and their contextual relationships, with a pre-constructed domain consensus semantic framework. This comparison aims to assess whether the logical roles and semantic relationships of the extracted technical terms or abbreviations conform to general knowledge and norms within the domain. For example, if the usage of a term in the extracted key information differs significantly from its typical usage in the domain consensus framework, or if other related information logically contradicts it, an inconsistency risk is considered. Furthermore, it assesses whether the expressed meaning deviates from general knowledge in the domain; for example, whether a term has been misused or given a non-standard meaning in a specific context.

[0084] Based on the results of deep semantic context consistency comparison, the certainty of the information extraction process is adjusted. When there is a risk of inconsistency in the deep semantic context, the certainty of information extraction is reduced, and a risk warning containing specific contradictions or deviations is generated. This means that if the comparison finds that the extracted key information is semantically or logically inconsistent with the domain consensus framework, the system will consider the extraction process to have a high degree of uncertainty, thus reducing its certainty. Simultaneously, the system will generate a detailed risk warning, clearly pointing out specific contradictions or deviations, such as the term 'XXX' being used as a subject here, which is inconsistent with its common use as an object in the domain consensus, or the description of 'YYY' having a logical conflict with industry standard Z, so that problems can be quickly located during subsequent manual verification.

[0085] This application addresses the problem that information extraction methods may lead to inaccurate assessments of information certainty due to misunderstandings of technical terms or abbreviations, contextual deviations, or logical inconsistencies by introducing a deep semantic context consistency comparison mechanism.

[0086] In some embodiments, the step of comparing the extracted key information with other relevant information of the supplier and industry reference information to obtain the degree of consistency between the extracted key information and other relevant information and industry reference information includes:

[0087] A quality assessment is performed on the industry reference information used for comparison. The quality assessment includes:

[0088] Check the generation or update time of the reference information to determine whether the generation or update time is within the valid time window of the current bidding project, and obtain the timeliness assessment result of the reference information.

[0089] Analyze the business scope, technical field, or project scale covered by the reference information, determine the degree of matching between the business scope, technical field, or project scale and the current bidding project or the extracted key information, and obtain the scope matching degree evaluation result of the reference information.

[0090] By comparing the statistical indicator definitions, units of measurement, or evaluation standards used in the reference information, it is determined whether the statistical indicator definitions, units of measurement, or evaluation standards are consistent with the current bidding project or the key information extracted, and the consistency evaluation result of the reference information is obtained.

[0091] Based on the timeliness assessment results, scope matching assessment results, and consistency assessment results, a reference information quality score is calculated for each piece of reference information;

[0092] When the quality score of the reference information is lower than the preset threshold, the weight of the reference information in the calculation of the degree of fit is reduced, or the reference information is marked as unreliable reference information.

[0093] By combining the reference information with reduced weights or marked as unreliable, the extracted key information is compared with other relevant information of the supplier and industry reference information to obtain the degree of consistency between the extracted key information and other relevant information and industry reference information.

[0094] Specifically, a quality assessment of the industry reference information used for comparison aims to ensure the high reliability and relevance of the reference data. This quality assessment can be broken down into several dimensions. First, by examining the generation or update time of the reference information, its timeliness can be determined. For example, if an industry report was published much earlier than the start time of the current bidding project, its timeliness assessment result may be low. Second, by analyzing the business scope, technical field, or project scale covered by the reference information, its degree of matching with the current bidding project or extracted key information can be assessed, thus obtaining a scope matching assessment result. For example, reference information about a large infrastructure project may not be applicable to assessing key information for a small software development project. Third, by comparing the statistical indicator definitions, units of measurement, or evaluation standards used in the reference information, it can be determined whether its scope is consistent with that of the current bidding project or extracted key information, thus obtaining a scope consistency assessment result. For example, if a reference information uses a compound annual growth rate while the key information uses an annual growth rate, their scope consistency needs to be assessed.

[0095] Based on the timeliness assessment, scope matching assessment, and consistency assessment results, a comprehensive reference information quality score can be calculated for each piece of reference information. This score can be a weighted average, where each assessment dimension is assigned a different weight according to its importance. For example, timeliness may be given a higher weight in some rapidly changing industries. When the calculated reference information quality score is lower than a preset threshold, it indicates that the quality of the reference information is problematic. In this case, two strategies can be adopted: first, reduce the weight of the reference information in the calculation of the degree of fit, so as to reduce its impact on the final degree of fit; second, directly mark the reference information as unreliable reference information, thereby excluding it or giving it a very low weight in subsequent comparisons. Finally, by combining the reference information after weight adjustment or labeling, the extracted key information is compared with other relevant information from the supplier and industry reference information to obtain a more accurate and reliable degree of fit.

[0096] This application effectively addresses the limitations of the accuracy calculation caused by the quality issues of reference information by introducing a quality assessment mechanism for industry reference information.

[0097] In some embodiments, the degree of certainty in the above-described process of evaluating the extraction of key information from electronic records, wherein the degree of certainty reflects whether the information extraction is a direct identification, specifically:

[0098] First, part-of-speech tagging and dependency analysis are performed on the words in the electronic archive text to identify key information entities and their modifiers, predicate verbs, and object structures. This yields the syntactic structure and part-of-speech combinations of the key information entities within the sentences. Part-of-speech tagging assigns a corresponding part-of-speech label to each word in the text, such as noun, verb, or adjective. Dependency analysis identifies the grammatical relationships between words in the sentences, such as subject-verb, verb-object, and modifier relationships. Through these analyses, the grammatical roles and structures of key information within sentences can be accurately located.

[0099] Secondly, based on the syntactic structure and part-of-speech combination of key information entities in the sentence, it is determined whether the key information entities conform to the preset structured information pattern. If they do, the key information entities are determined to be directly identifiable. The preset structured information pattern can be a series of well-defined grammatical rules or templates, such as company name + registered capital + numerical value, etc., the purpose of which is to identify key information presented in a clear and standardized form.

[0100] Furthermore, among the key information entities determined to be directly identifiable, the modifiers and predicate verbs of the key information entities are analyzed to identify whether there are any ambiguous or vague expressions. Specifically, ambiguous expressions may refer to unclear words, such as "several" or "parts"; ambiguous expressions refer to situations where a word or phrase may have multiple interpretations.

[0101] When ambiguous or vague expressions are identified, even if the syntactic structure of the key information entity conforms to the direct recognition pattern, the certainty of extracting the key information should be reduced. This is because ambiguous or vague expressions increase the uncertainty of information understanding and judgment, even if their surface structure seems clear.

[0102] Finally, based on the reduced level of certainty, the certainty assessment result of the key information is output. This assessment result will serve as an important input for subsequent calculations of the key information credibility metric.

[0103] This application achieves a deep understanding of how key information is extracted through detailed syntactic and semantic analysis of electronic archival text. By employing part-of-speech tagging and dependency parsing, the system can identify the precise structure and grammatical role of key information within the text, thereby determining whether it is presented directly in a structured format.

[0104] In some embodiments, the step of calculating the credibility metric of the extracted key information based on the storage medium's operational status information, integrity verification records, the degree of certainty of the processing procedure, and the degree of consistency includes:

[0105] A correlation analysis is performed on the operational status information of the storage medium, integrity verification records, the degree of certainty of the processing process, and the degree of consistency to identify the dependencies and redundant information between the evaluation dimensions.

[0106] Based on the correlation analysis results, a feature space that can reflect the independent contribution of each evaluation dimension is constructed, and the evaluation dimensions are reduced in dimensionality to remove redundant information.

[0107] In the feature space after dimensionality reduction, calculate the degree of independent influence of each evaluation dimension on the reliability of key information;

[0108] Based on the degree of independent influence, assign corresponding weights to each evaluation dimension;

[0109] Based on the assigned weights, and in conjunction with the storage medium's operational status information, integrity verification records, the degree of certainty of the processing procedure, and the degree of consistency, a reliability metric for the extracted key information is calculated.

[0110] Specifically, correlation analysis refers to using statistical methods or machine learning algorithms, such as Pearson correlation coefficient, mutual information, principal component analysis, or factor analysis, to quantify the relationships among four evaluation dimensions: the operational status information of the storage medium, integrity verification records, the certainty of the processing, and the degree of conformity. Its purpose is to reveal whether these dimensions exhibit positive correlations, negative correlations, or more complex non-linear dependencies, and to identify any redundant information, such as information overlap or repeated expressions. For example, poor operational status of the storage medium may directly lead to problems with integrity verification records, indicating a dependency and potential redundancy between the two.

[0111] The construction of a feature space and dimensionality reduction can be understood as follows: after identifying the dependencies and redundant information between evaluation dimensions, mathematical transformations are used to map the original, potentially interconnected evaluation dimensions to a new, lower-dimensional feature space where each dimension is independent. Specific methods for dimensionality reduction can include principal component analysis, linear discriminant analysis, or independent component analysis. The aim is to eliminate redundant information in the original data and extract the core features that best represent the reliability of key information, thereby ensuring the accuracy and efficiency of subsequent calculations.

[0112] Calculating the independent influence of each evaluation dimension on the reliability of key information in the dimensionality-reduced feature space refers to quantifying the contribution of each independent feature to the final credibility of the key information after eliminating inter-dimensional dependencies and redundancy. This can be achieved through methods such as regression analysis, decision tree models, or support vector machines to determine the influence weight of each independent feature on the credibility metric. The goal is to ensure that each dimension reflects its true and independent contribution in the final credibility calculation.

[0113] Assigning weights to each assessment dimension based on its degree of independent influence means assigning an importance coefficient to each dimension in the final credibility metric calculation based on the magnitude of its independent impact on the reliability of key information. Dimensions with greater influence receive higher weights. The aim is to ensure that the final credibility metric more accurately reflects the true reliability level of key information across different dimensions.

[0114] Based on the assigned weights and combined with the storage medium's operating status information, integrity verification records, the certainty of the processing process, and the degree of consistency, the credibility metric of the extracted key information is calculated. This means that the various evaluation dimensions after weight allocation are comprehensively calculated according to a preset calculation model (such as weighted average, multi-factor product model, or more complex machine learning model) to obtain the final credibility metric of the key information.

[0115] This application first identifies potential dependencies and redundant information among four evaluation dimensions—the operational status information of the storage medium, integrity verification records, the certainty of the processing procedure, and the degree of consistency—through correlation analysis. These dependencies and redundancy could lead to biased evaluation results if calculated directly. By constructing a feature space reflecting the independent contribution of each evaluation dimension and performing dimensionality reduction, redundant information is effectively removed, ensuring that the features used in subsequent calculations are independent and representative. Based on this, the independent influence of each evaluation dimension on the reliability of key information is calculated in the dimensionality-reduced feature space, quantifying the true contribution of each dimension in the credibility assessment. Finally, weights are assigned to each evaluation dimension based on these independent influences, and a weighted calculation is performed using the original evaluation information to obtain a more accurate, objective, and interpretable measure of the credibility of key information. This method avoids information distortion that may result from simple aggregation, ensuring the scientific validity and reliability of the evaluation results.

[0116] In some embodiments, the step of generating an identification information containing key information, a credibility measurement value, the reason why the credibility measurement value is lower than the preset judgment standard, and a suggested scope of verification when the credibility measurement value is lower than the preset judgment standard includes:

[0117] Identify the correlation between multiple pieces of key information with credibility below a preset judgment standard. The correlation includes logical dependence, content overlap, or causal relationship.

[0118] Based on the identified correlations, construct a correlation structure that describes the relationships between key information with credibility levels lower than the preset judgment criteria;

[0119] Based on the association structure, key information with credibility below the preset judgment standard is grouped, and key information with strong association is grouped together.

[0120] Generate a comprehensive verification prompt for each information group. The comprehensive verification prompt includes the identification information of all key information in the group whose credibility is lower than the preset judgment standard, their respective credibility measurement value, the reason why their respective credibility measurement value is lower than the preset judgment standard, the suggested scope of verification, and a description of the correlation between the key information in the information group.

[0121] Provide an interactive interface to display comprehensive verification prompts and allow manual visualization and interactive exploration of related structures.

[0122] Specifically, identifying the correlation between multiple pieces of key information with credibility levels below a preset judgment standard refers to the system using natural language processing, semantic analysis, knowledge graph technology, or preset domain rules to analyze the interrelationships of these key pieces of information in terms of content, context, logical relationships, or business processes. For example, if a supplier's financial statement has low credibility for a certain data point, and the funding amount mentioned in its project experience description contradicts the financial data, then there is a logical dependency or content overlap between these two pieces of key information. Correlation can be understood as the mutual influence or common direction between information in terms of semantics, structure, or business logic.

[0123] Based on the identified correlations, a relational structure is constructed to describe the relationships between key information with credibility levels below a preset threshold. This can be understood as representing these low-credibility key information pieces and their identified correlations in graph, network, or matrix form. For example, a knowledge graph can be constructed where nodes represent key information, edges represent the types of relationships between them (such as influence, inclusion, contradiction, etc.), and can include correlation strength or confidence level. The aim is to integrate scattered risk points into a structured, easily analyzable overall view.

[0124] Based on the association structure, key information with credibility below a preset judgment standard is grouped together, grouping key information with strong correlations together. Strong correlation can be defined by a preset threshold or clustering algorithm. For example, if the correlation between two key information pieces exceeds a certain threshold, or if they form a tight cluster in the association structure, they are considered to be strongly correlated. The purpose of grouping is to centrally process interrelated risk points, avoiding overlooking their inherent connections during manual verification.

[0125] A comprehensive verification prompt is generated for each information group. This prompt not only includes identification information for all low-credibility key information within the group, their respective credibility metrics, the reasons why the credibility metrics are lower than the preset judgment standard, and the suggested scope of verification, but also adds a description of the correlation between the key information within the information group. For example, the prompt may explicitly point out that there is a discrepancy between the revenue data in Supplier A's financial statements and the contract amount in the project experience, thus providing human verification with more comprehensive background information and verification direction.

[0126] An interactive interface is provided to display comprehensive verification prompts and allow manual viewing and interactive exploration of the interconnected structure. This interface can graphically display the interconnected structure, such as through a node-edge graph. Users can click on nodes to view detailed information about key points, drag and zoom to explore local or overall views of the interconnected network, and even perform preliminary editing or annotation of the interconnected structure to aid understanding and decision-making. Its purpose is to present complex interconnected information intuitively, lower the barrier to human understanding, and support in-depth analysis.

[0127] This application identifies the correlations between multiple key pieces of information with credibility levels below a preset judgment standard, constructs a correlation structure based on this, and then groups these key pieces of information, thereby integrating originally scattered and independent verification points into logically connected information groups. This integration mechanism allows human operators to receive verification prompts not only without isolated information fragments, but also to obtain a comprehensive view containing multiple related risk points and their interrelationships. By generating comprehensive verification prompts for each information group and clearly indicating the correlation descriptions between key information within the group, the system can guide human operators to grasp the overall picture of the problem from a macro perspective, avoiding misunderstandings or verification omissions caused by fragmented information.

[0128] In some embodiments, the step of providing an interactive interface for receiving the results of manual review of the original electronic file indicated by the verification prompt, and updating the final credibility status of key information based on the review results, includes:

[0129] Syntactic analysis is performed on the text content in the original electronic archives to identify key information entities, modifiers, predicate verbs and object structures, and their semantic relationships are extracted;

[0130] Content recognition is performed on the image information in the original electronic archives to extract the visual elements of text, charts, and seals contained in the images, and the correspondence between these visual elements and the text content is analyzed.

[0131] Based on the syntactic analysis results and content recognition results, a semantic association graph describing the key information and their interrelationships in the original electronic archives is constructed;

[0132] The interactive interface simultaneously presents the text content, image information, and semantic association graph of the original electronic archives.

[0133] The interactive interface allows users to manually re-annotate, correct, or add new relationships to key information entities in the semantic association graph by selecting, dragging, or clicking.

[0134] The interactive interface displays in real time the potential impact of human intervention on the credibility status of key information;

[0135] Based on the manually corrected key information and relationships, the system recalculates the credibility metric of the key information;

[0136] Update the final credibility status of key information based on the recalculated credibility metric.

[0137] Specifically, syntactic analysis of the text content in original electronic files refers to using natural language processing (NLP) technology to perform part-of-speech tagging and dependency parsing on the textual information in the electronic files. This identifies key information entities in the text, such as supplier names, project amounts, and delivery dates, as well as modifiers, predicate verbs, and object structures between these entities, and then extracts the semantic relationships between them. For example, it can be identified that supplier A promises to complete the project within 30 days; supplier A is the subject, the promise is the predicate, and completing the project within 30 days is the object, thus extracting the promise relationship between supplier A and project completion. Content recognition of image information in original electronic files refers to using technologies such as Optical Character Recognition (OCR) and image analysis to extract visual elements such as text, charts, and seals from the image portion of the electronic files. For example, it can identify contract terms in scanned documents, numerical charts in financial statements, and images of official seals. Simultaneously, the correspondence between these visual elements and the text content is analyzed; for example, it can be identified that Appendix 1 mentioned in the text is related to a certain chart in the image.

[0138] Based on syntactic analysis and content recognition results, a semantic association graph describing key information and their interrelationships in the original electronic archives is constructed. The aim is to present the key information and its inherent connections scattered throughout the text and images in a structured graphical format. This graph can intuitively display various key information entities (such as suppliers, projects, amounts, dates, qualifications, etc.) and their semantic relationships (such as belonging, inclusion, commitment, signing, etc.), forming a knowledge network. Furthermore, the text content, image information, and semantic association graph of the original electronic archives are simultaneously presented in the interactive interface, providing a comprehensive and multi-dimensional view for manual review. Humans can compare the original text and images with the semantic association graph to more accurately understand the context and logical relationships of the information. As a preferred implementation, the interactive interface allows humans to re-annotate, correct, or add new relationships to key information entities in the semantic association graph through selection, dragging, or clicking. For example, if the system automatically identifies an entity inaccurately, the human can directly click on the entity in the graph to edit it; if the system fails to identify an important relationship, the human can drag and drop two entities to establish a new relationship. Its purpose is to endow human operators with a high degree of flexibility and accuracy to correct biases in machine recognition or supplement missing information. Furthermore, the interactive interface displays in real-time the potential impact of human operations on the credibility status of key information, aiming to provide immediate feedback. For example, when a human corrects a key information entity or adds a new relationship, the system immediately reassesses the credibility of the key information based on these modifications, displaying the new credibility metric or its changing trend to the human operator, thus assisting in judging the effectiveness of the correction. Therefore, based on the manually corrected key information and relationships, the system recalculates the credibility metric of the key information and updates the final credibility status of the key information based on the recalculated credibility metric. This process ensures that the results of human review are reflected in the final credibility assessment of key information in a timely and accurate manner, thereby improving the credibility of information throughout the entire bidding process.

[0139] This application effectively solves the problems of low efficiency and insufficient accuracy that may be faced by manual review in the basic scheme by constructing and displaying the semantic association graph of the original electronic archives and providing a highly interactive review mechanism.

[0140] In some embodiments, the step of providing an interactive interface for receiving the results of manual review of the original electronic file indicated by the verification prompt, and updating the final credibility status of key information based on the review results, includes:

[0141] Receive manual corrections of key information in the original electronic files indicated by the verification prompts;

[0142] A consistency check is performed on the revised key information. The consistency check includes:

[0143] The corrected key information is compared with the existing storage risk information generated by the storage medium operation status assessment module to determine whether the correction contradicts the storage risk information.

[0144] The corrected key information is compared with the information extraction uncertainty information already in the system and generated by the information extraction certainty assessment module to determine whether the correction contradicts the information extraction uncertainty information.

[0145] When the corrected key information contradicts the stored risk information or the information extraction uncertainty information, a contradiction prompt is generated. The contradiction prompt includes the corrected key information, the type of contradiction, the risk information involved in the contradiction, and the scope of recommended manual re-verification.

[0146] The interactive interface displays conflicting prompts in real time and allows manual confirmation or correction of these prompts.

[0147] Based on the key information that has been manually confirmed or revised, the system recalculates the credibility metric of the key information.

[0148] Update the final credibility status of key information based on the recalculated credibility metric.

[0149] Specifically, the operation of receiving manual corrections of key information in the original electronic files indicated by the verification prompts refers to the ability of an operator, within the interactive interface, to modify, supplement, or delete key information in the original electronic files that is identified as having low credibility, based on the verification prompts. For example, an operator can correct an incorrect amount or add a missing supplier name.

[0150] The purpose of performing a consistency check on the revised key information is to ensure the reasonableness and accuracy of the manual corrections and to avoid introducing new data inconsistencies. Specifically, this consistency check includes two main aspects:

[0151] The corrected key information is compared with existing storage risk information generated by the storage medium operational status assessment module. Storage risk information reflects the health status of the electronic archive storage medium and the risk to data integrity. For example, if an electronic archive's storage medium is marked as high-risk (e.g., frequent input and output errors, corrupted data records), but key information extracted from the archive is manually corrected to be completely reliable, the system will identify this contradiction.

[0152] The corrected key information is compared with the information extraction uncertainty information already generated by the information extraction certainty assessment module. Information extraction uncertainty information reflects the difficulty and potential ambiguity in automatically extracting key information from electronic documents. For example, if a key piece of information is assessed by the system as having low extraction certainty (e.g., the original text is vague or ambiguous), and it is manually corrected to a clear and unambiguous value, the system will check whether this correction is reasonable and whether it contradicts the original extraction uncertainty.

[0153] When the corrected key information contradicts storage risk information or information retrieval uncertainty information, the system generates a contradiction alert. This alert aims to clearly inform users of potential problems with manual correction and provide necessary contextual information. The alert includes the corrected key information itself, the specific type of contradiction identified (e.g., contradiction with storage risk or retrieval uncertainty), the specific risk information involved (e.g., bad block risk on the storage medium or ambiguity in the original text), and the scope of suggested manual re-verification (e.g., please re-examine the storage status of the original file or carefully check the original text context).

[0154] The interactive interface displays conflicting prompts in real time and allows manual confirmation or correction. This means that after a manual correction, the system immediately reports the potential conflict, and the user can choose to accept the conflict (e.g., if the user has a higher level of judgment) or correct the key information again based on the prompts to eliminate the conflict.

[0155] Based on the key information that has been manually verified or revised, the system recalculates the credibility metric of the key information. This recalculation process comprehensively considers the manually revised information as well as the system's internal risk assessment results, ensuring that the final credibility metric more accurately reflects the true reliability of the key information. Finally, based on the recalculated credibility metric, the system updates the final credibility status of the key information and displays it through the interactive interface.

[0156] This application effectively solves the problem that manual review results may introduce new inconsistencies or errors in the basic scheme by introducing a consistency check mechanism for manual correction operations.

[0157] like Figure 2 As shown in the embodiments, this application also discloses an intelligent bidding process management system assisted by electronic records, including:

[0158] Storage location information acquisition module 1 is used to acquire the storage location information of electronic archives;

[0159] The operation status information acquisition module 2 is used to acquire the operation status information of the storage medium of the electronic archive and the integrity verification record of the electronic archive based on the storage location information;

[0160] Certainty assessment module 3 is used to assess the degree of certainty in the process of extracting key information from electronic archives;

[0161] The matching degree comparison module 4 is used to compare the extracted key information with other relevant information of the supplier and industry reference information to obtain the matching degree between the extracted key information and other relevant information and industry reference information.

[0162] The trust metric calculation module 5 is used to calculate the trust metric of the extracted key information based on the operating status information of the storage medium, integrity verification records, the degree of certainty of the processing process, and the degree of consistency.

[0163] The verification prompt generation module 6 is used to generate a verification prompt containing key information, the credibility value, the reason why the credibility value is lower than the preset judgment standard, and the suggested verification scope when the credibility measurement value is lower than the preset judgment standard.

[0164] The bid evaluation report module 7 is used to integrate verification prompts into the bid evaluation report, thereby transforming the underlying data uncertainties that machines cannot resolve independently into human understanding and operational verification.

[0165] The review result update module 8 is used to provide an interactive interface for receiving the results of manual review of the original electronic files pointed to by the review prompt, and to update the final credibility status of key information based on the review results.

[0166] Display module 9 is used to display the final credibility status of key information through an interactive interface, serving as auxiliary information for human decision-making.

[0167] The electronic record-assisted intelligent bidding process management system of this application can effectively solve the problem that traditional intelligent bidding systems suffer from decision-making errors and difficulty in tracing due to data concealment bias when processing electronic records stored in sub-healthy storage devices.

[0168] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for managing intelligent bidding processes assisted by electronic records, characterized in that, include: Obtain the storage location information of electronic records; Based on the storage location information, obtain the operational status information of the storage medium of the electronic archives and the integrity verification records of the electronic archives; Assess the degree of certainty in the process of extracting key information from electronic records; The extracted key information is compared with other relevant information of the supplier and industry reference information to obtain the degree of consistency between the extracted key information and other relevant information of the supplier and industry reference information. Based on the storage medium's operational status information, integrity verification records, the degree of certainty and consistency of the processing procedure, calculate the credibility metric of the extracted key information; When the credibility metric value is lower than the preset judgment standard, a verification prompt is generated, which includes key information, the credibility metric value, the reason why the credibility metric value is lower than the preset judgment standard, and the suggested scope of verification. By incorporating verification prompts into the bid evaluation report, the underlying data uncertainties that machines cannot resolve independently can be transformed into human understanding and operational verification. Provide an interactive interface to receive the results of manual review of the original electronic files indicated by the verification prompts, and update the final credibility status of key information based on the review results; The final credibility status of key information is displayed through an interactive interface to serve as auxiliary information for human decision-making.

2. The method for intelligent bidding process management assisted by electronic archives according to claim 1, characterized in that, Also includes: At the controller level of the storage array, capture micro-behavioral data of every input and output operation; For each storage cell in the storage array, establish a normal behavioral fingerprint describing the distribution of its microscopic input and output behavioral characteristics under different load conditions; When the captured micro-behavioral data of input and output operations deviates statistically from the normal behavioral fingerprint of the storage unit, the data block involved in the operation is marked as transiently high risk; Adjust the source reliability factor of electronic records based on the instantaneous high-risk markers; The adjusted source reliability factor is applied to the calculation of the credibility metric value of the extracted key information.

3. The method for intelligent bidding process management assisted by electronic archives according to claim 1, characterized in that, The steps for assessing the certainty of the process of extracting key information from electronic records include: Identify technical terms or abbreviations in key information extracted from electronic archives, and analyze the logical roles and semantic relationships of these terms or abbreviations in the current sentences, paragraphs, and documents. Based on all the bidding documents in the current batch, construct an association network consisting of high-frequency professional terms or abbreviations and their contextual roles, as a domain consensus semantic framework. The extracted key information is compared with the domain consensus semantic framework in a deep semantic context consistency comparison. The logical roles and semantic relationships of professional terms or abbreviations in the key information are evaluated to determine whether they are consistent with the domain consensus semantic framework, whether the related information is logically reasonable, and whether the meaning expressed deviates from the general understanding of the domain. Based on the results of deep semantic context consistency comparison, the degree of certainty in the information extraction process is adjusted, specifically as follows: When there is a risk of inconsistency in the deep semantic context, the certainty of information extraction is reduced, and risk warnings containing specific contradictions or deviations are generated.

4. The method for intelligent bidding process management assisted by electronic archives according to claim 1, characterized in that, The step of comparing the extracted key information with other relevant information from the supplier and industry reference information to determine the degree of consistency between the extracted key information and other relevant information and industry reference information includes: A quality assessment is performed on the industry reference information used for comparison. The quality assessment includes: Check the generation or update time of the reference information to determine whether the generation or update time is within the valid time window of the current bidding project, and obtain the timeliness assessment result of the reference information. Analyze the business scope, technical field, or project scale covered by the reference information, determine the degree of matching between the business scope, technical field, or project scale and the current bidding project or the extracted key information, and obtain the scope matching degree evaluation result of the reference information. By comparing the statistical indicator definitions, units of measurement, or evaluation standards used in the reference information, it is determined whether the statistical indicator definitions, units of measurement, or evaluation standards are consistent with the current bidding project or the key information extracted, and the consistency evaluation result of the reference information is obtained. Based on the timeliness assessment results, scope matching assessment results, and consistency assessment results, a reference information quality score is calculated for each piece of reference information; When the quality score of the reference information is lower than the preset threshold, the weight of the reference information in the calculation of the degree of fit is reduced, or the reference information is marked as unreliable reference information. By combining the reference information with reduced weights or marked as unreliable, the extracted key information is compared with other relevant information of the supplier and industry reference information to obtain the degree of consistency between the extracted key information and other relevant information and industry reference information.

5. The method for intelligent bidding process management assisted by electronic archives according to claim 1, characterized in that, The assessment evaluates the degree of certainty in the process of extracting key information from electronic records, wherein the degree of certainty reflects whether the information extraction is direct identification, specifically: Part-of-speech tagging and dependency analysis are performed on words in electronic archive texts to identify key information entities and their modifiers, predicate verbs and object structures, and to obtain the syntactic structure and part-of-speech combination of key information entities in sentences; Based on the syntactic structure and part-of-speech combination of key information entities in the sentence, it is determined whether the key information entities conform to the preset structured information pattern. If they do, the key information entities are determined to be directly identifiable. Among the key information entities that are determined to be directly identifiable, analyze the modifiers and predicate verbs of the key information entities to identify whether there are any ambiguous or vague expressions. When ambiguous or vague expressions are identified, the certainty of extracting key information is reduced; Based on the reduced level of certainty, output the assessment results of the certainty level of key information.

6. The method for intelligent bidding process management assisted by electronic archives according to claim 1, characterized in that, The step of calculating the credibility metric of the extracted key information based on the storage medium's operating status information, integrity verification records, the certainty of the processing procedure, and the degree of consistency includes: A correlation analysis is performed on the operational status information of the storage medium, integrity verification records, the degree of certainty of the processing process, and the degree of consistency to identify the dependencies and redundant information between the evaluation dimensions. Based on the correlation analysis results, a feature space that can reflect the independent contribution of each evaluation dimension is constructed, and the evaluation dimensions are reduced in dimensionality to remove redundant information. In the feature space after dimensionality reduction, calculate the degree of independent influence of each evaluation dimension on the reliability of key information; Based on the degree of independent influence, assign corresponding weights to each evaluation dimension; Based on the assigned weights, and in conjunction with the storage medium's operational status information, integrity verification records, the degree of certainty of the processing procedure, and the degree of consistency, a reliability metric for the extracted key information is calculated.

7. The method for intelligent bidding process management assisted by electronic archives according to claim 1, characterized in that, The steps for generating a verification prompt containing key information, the credibility metric, the reason why the credibility metric is lower than the preset judgment standard, and a suggested scope of verification when the credibility metric is lower than the preset judgment standard include: Identify the correlation between multiple pieces of key information with credibility below a preset judgment standard. The correlation includes logical dependence, content overlap, or causal relationship. Based on the identified correlations, construct a correlation structure that describes the relationships between key information with credibility levels lower than the preset judgment criteria; Based on the association structure, key information with credibility below the preset judgment standard is grouped, and key information with strong association is grouped together. Generate a comprehensive verification prompt for each information group. The comprehensive verification prompt includes the identification information of all key information in the group whose credibility is lower than the preset judgment standard, their respective credibility measurement value, the reason why their respective credibility measurement value is lower than the preset judgment standard, the suggested scope of verification, and a description of the correlation between the key information in the information group. Provide an interactive interface to display comprehensive verification prompts and allow manual visualization and interactive exploration of related structures.

8. The method for intelligent bidding process management assisted by electronic archives according to claim 1, characterized in that, The step of providing an interactive interface to receive the results of manual review of the original electronic file indicated by the verification prompt, and updating the final credibility status of key information based on the review results, includes: Syntactic analysis is performed on the text content in the original electronic archives to identify key information entities, modifiers, predicate verbs and object structures, and their semantic relationships are extracted; Content recognition is performed on the image information in the original electronic archives to extract the visual elements of text, charts, and seals contained in the images, and the correspondence between these visual elements and the text content is analyzed. Based on the syntactic analysis results and content recognition results, a semantic association graph describing the key information and their interrelationships in the original electronic archives is constructed; The interactive interface simultaneously presents the text content, image information, and semantic association graph of the original electronic archives. The interactive interface allows users to manually re-annotate, correct, or add new relationships to key information entities in the semantic association graph by selecting, dragging, or clicking. The interactive interface displays in real time the potential impact of human intervention on the credibility status of key information; Based on the manually corrected key information and relationships, the system recalculates the credibility metric of the key information; Update the final credibility status of key information based on the recalculated credibility metric.

9. The method for intelligent bidding process management assisted by electronic archives according to claim 1, characterized in that, The step of providing an interactive interface to receive the results of manual review of the original electronic file indicated by the verification prompt, and updating the final credibility status of key information based on the review results, includes: Receive manual corrections of key information in the original electronic files indicated by the verification prompts; A consistency check is performed on the revised key information. The consistency check includes: The corrected key information is compared with the existing storage risk information generated by the storage medium operation status assessment module to determine whether the correction contradicts the storage risk information. The corrected key information is compared with the information extraction uncertainty information already in the system and generated by the information extraction certainty assessment module to determine whether the correction contradicts the information extraction uncertainty information. When the corrected key information contradicts the stored risk information or the information extraction uncertainty information, a contradiction prompt is generated. The contradiction prompt includes the corrected key information, the type of contradiction, the risk information involved in the contradiction, and the scope of recommended manual re-verification. The interactive interface displays conflicting prompts in real time and allows manual confirmation or correction of these prompts. Based on the key information that has been manually confirmed or revised, the system recalculates the credibility metric of the key information. Update the final credibility status of key information based on the recalculated credibility metric.

10. An intelligent bidding process management system assisted by electronic records, characterized in that, The system includes: The storage location information acquisition module is used to acquire the storage location information of electronic archives; The operation status information acquisition module is used to acquire the operation status information of the storage medium of the electronic archive and the integrity verification record of the electronic archive based on the storage location information; The certainty assessment module is used to assess the degree of certainty in the process of extracting key information from electronic archives; The matching degree comparison module is used to compare the extracted key information with other relevant information of the supplier and industry reference information to obtain the degree of matching between the extracted key information and other relevant information and industry reference information. The trust metric calculation module is used to calculate the trust metric of the extracted key information based on the storage medium's operating status information, integrity verification records, the degree of certainty of the processing process, and the degree of consistency. The verification prompt generation module is used to generate a verification prompt containing key information, the credibility value, the reason why the credibility value is lower than the preset judgment standard, and the suggested verification scope when the credibility metric value is lower than the preset judgment standard. The bid evaluation report module is used to integrate verification prompts into the bid evaluation report, thereby transforming the underlying data uncertainties that machines cannot resolve independently into human understanding and operational verification. The review result update module provides an interactive interface for receiving the results of manual review of the original electronic files indicated by the review prompts, and updates the final credibility status of key information based on the review results. The display module is used to show the final credibility status of key information through an interactive interface, serving as auxiliary information for human decision-making.

Images