Intelligent bid evaluation method based on big data

Abstract

The invention provides an intelligent bid evaluation method based on big data, and belongs to the technical field of big data. Comprising the steps of 1, collecting original data related to a bid invitation project, and identifying and extracting key index information in the original data by using a natural language processing technology according to a preset field; 2, preprocessing the key index information to obtain standard index information, and storing the standard index information in a data warehouse; 3, substituting the standard index information of the data warehouse and the corresponding index field into the weight model to obtain a first weight; 4, taking the historical bidding and tendering data of the data warehouse as a training set, taking the standard bidding file as a test set, and training the latest index parameters of the bid evaluation large model by using the training set to obtain a final bid evaluation large model; and 5, forming a basic bid evaluation brain of the big data model by the final bid evaluation big model, and substituting the actual bid invitation file into the system to obtain a bid evaluation report result. Intelligent bid evaluation management of the whole process is achieved, and bid evaluation efficiency and fairness are effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of big data technology, and in particular to an intelligent bid evaluation method based on big data. Background Art

[0002] At present, there are many intelligent bid evaluation technologies that have been implemented in the domestic electronic bidding industry, such as anonymous bidding evaluation system, intelligent auxiliary bid evaluation, electronic bidding system, etc., which cannot realize the full process supervision of the bidding process, the evaluation workload is large and the efficiency is low, the evaluation results lack objectivity and fairness, and the data accuracy and legality cannot be verified in real time. Comprehensive intelligent bid evaluation technology is a major barrier to my country's electronic bidding industry, which has increased a lot of manpower and costs for bidding companies. So far, there has been no relevant technological breakthrough in China.

[0003] Therefore, the present invention proposes an intelligent bid evaluation method based on big data. Summary of the invention

[0004] The present invention provides an intelligent bid evaluation method based on big data, which reduces the impact of human factors on bid evaluation results by utilizing big data analysis and artificial intelligence technology, combining the collection, organization and analysis of a large amount of bidding data, and simplifies the bid evaluation process through automated tools.

[0005] In one aspect, the present invention provides an intelligent bid evaluation method based on big data, comprising:

[0006] Step 1: Collect raw data related to the bidding project, and use natural language processing technology to identify and extract key indicator information from the raw data according to preset fields;

[0007] Step 2: Preprocess the key indicator information to obtain standard indicator information and store it in the data warehouse;

[0008] Step 3: Substitute the standard indicator information and corresponding indicator fields of the data warehouse into the weight model to obtain the first weights corresponding to different indicator fields;

[0009] Step 4: Use the historical bidding data in the data warehouse as the training set and the standard bidding documents as the test set. Use the training set to train the bid evaluation model. Use the test set and the first weights of different indicator fields to update the latest indicator parameters of the bid evaluation model to obtain the final bid evaluation model.

[0010] Step 5: Build the final bid evaluation model into the bid evaluation brain system to form the bid evaluation brain based on the big data model. Substitute the actual bidding documents into the bid evaluation brain system to obtain the bid evaluation report results.

[0011] On the other hand, collect original data related to the bidding project, including:

[0012] Introduce blockchain technology, design blockchain network architecture, and preset node permissions and data sharing mechanisms of blockchain network;

[0013] Determine the main data source for bidding and develop the interface of the main data source according to the preset interface standards;

[0014] The original interface data of the bidding project is obtained through the blockchain subject market data interface, and the original data of the bidding project is obtained through data format conversion preprocessing operations, wherein the original data includes formatted data and unformatted data.

[0015] On the other hand, natural language processing technology is used to identify and extract key indicator information from the original data according to preset fields, including:

[0016] If the original data is formatted data, the corresponding field indicators in the original data are extracted using the field template and the field indicators are matched with the corresponding data results to generate the key indicator information of the field indicator-indicator data;

[0017] If the original data is unformatted data, the unformatted data is processed using an OCR tool, feature processing is performed on the unformatted data, useless data is filtered out, and a first feature sequence is obtained;

[0018] Match any unit feature of the first feature sequence with a standard dictionary, obtain the character with the highest matching degree as the unit character, and obtain text character data corresponding to the first feature sequence;

[0019] Using NER technology to identify entities of the text character data in combination with preset fields, and marking the entities according to the field type of the preset fields;

[0020] According to the annotation information of the entity, the entity is classified to obtain the key indicator information of the original data.

[0021] On the other hand, the key indicator information is preprocessed to obtain standard indicator information, which is then stored in the data warehouse, including:

[0022] The key indicator information is filtered to remove abnormal data, and then pre-processed according to the preset standard unit to obtain standard indicator information data;

[0023] Design the data table structure according to the preset standard fields and successfully match the data table with the data warehouse;

[0024] The standard indicator information data is stored in batches in the data table, and the data that does not conform to the data table fields is filtered into the junk table as useless data;

[0025] Complete the data storage of the data warehouse and start it using the SQL script.

[0026] On the other hand, the standard indicator information and corresponding indicator fields of the data warehouse are substituted into the weight model to obtain the first weights corresponding to different indicator fields, including:

[0027] Obtain the preset standard fields of the data warehouse and design the basic information of the scoring table; then combine any indicator field of the standard indicator information with the scoring criteria in the scoring table for evaluation to obtain the first score;

[0028] Where A1 represents the first score of the indicator field, j represents the jth scoring standard, m represents that the scoring table has m scoring standards in total, i represents the i-th indicator data value of the indicator field, n represents that the field has n indicator data values ​​in total, and L ij represents the logical length difference between the i-th indicator data value and the j-th scoring standard, σ1 represents the logical length difference between the i-th indicator data value and the j-th scoring standard, and σ2 represents the logical length difference between the i-th indicator data value and the j-th scoring standard. ij The variance of L j represents the logical length of the jth scoring criterion, f() represents the hierarchical superposition function, e represents the error coefficient, and w ij represents the hierarchical distance difference between the i-th indicator data value and the j-th scoring standard, w max Indicates the maximum length of the layer, w ave represents the maximum mean of the level;

[0029] Finally, the first scores of all indicators of the standard indicator information are obtained.

[0030] On the other hand, obtaining the first weights corresponding to different indicator fields also includes:

[0031] Substitute the first scores of all indicator fields of the standard indicator information and their corresponding indicator fields into the weight model;

[0032] Wherein, g represents the first weight of the indicator field, n1 represents a total of n1 indicator fields, σ i1 represents the initial evaluation weight of the i1th indicator field, A i1 represents the first score of the i1th indicator field, and lg() represents the weight model;

[0033] The first weights corresponding to different indicator fields of the standard indicator information are obtained.

[0034] On the other hand, the historical bidding data in the data warehouse is used as the training set, and the standard bidding documents are used as the test set. The training set is used to train the bid evaluation model, and the test set and the first weights of different indicator fields are used to update the latest indicator parameters of the bid evaluation model to obtain the final bid evaluation model, including:

[0035] Extract a data set containing bidding history records from the data warehouse as a training set, clean and filter the training set data, and then extract features from the training set to obtain different feature fields and their corresponding data;

[0036] Using the characteristic fields to perform a first training on the large bid evaluation model to obtain a first model indicator;

[0037] Afterwards, the large training and bid evaluation model is iteratively processed to obtain the model index corresponding to the number of iterations, and the model index and the first model index constitute the model index set;

[0038] The first standard parameter of the large bid evaluation model is obtained by using the model indicator set, and the standard bidding document is used as a test set for testing and evaluation to obtain the evaluation indicator of the large bid evaluation model;

[0039] According to the evaluation index and the first weights of different index fields, the first standard parameters of the large bid evaluation model are adjusted to obtain the hyper parameters of the large bid evaluation model;

[0040] After the hyperparameters are used to update the bid evaluation model, the final bid evaluation model is obtained.

[0041] On the other hand, the final bid evaluation model is built into the bid evaluation brain system to form the bid evaluation brain based on the big data model. The actual bidding documents are substituted into the bid evaluation brain system to obtain the bid evaluation report results, including:

[0042] Build the final bid evaluation model into the bid evaluation brain system to form the bid evaluation brain based on the big data model;

[0043] Create an API interface in the bid evaluation brain system and connect it to the front-end interface; upload the actual bidding documents through the front-end interface, generate initial bid evaluation results through the final bid evaluation model processing, use the business integration system to integrate the initial bid evaluation results into business, obtain the bid evaluation report results and output them to the front-end, and save the initial bid evaluation results in the back-end database.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] The present invention provides an intelligent bid evaluation method based on big data, which reduces the impact of human factors on bid evaluation results by utilizing big data analysis and artificial intelligence technology, combining the collection, organization and analysis of a large amount of bidding data, and simplifies the bid evaluation process through automated tools. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0047] Figure 1 It is a flowchart of an intelligent bid evaluation method based on big data provided by an embodiment of the present invention;

[0048] Figure 2 It is a structural diagram of the intelligent review system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] Embodiment 1:

[0051] like Figure 1 As shown, an intelligent bid evaluation method based on big data is provided in an embodiment of the present invention, comprising:

[0052] Step 1: Collect raw data related to the bidding project, and use natural language processing technology to identify and extract key indicator information from the raw data according to preset fields;

[0053] Step 2: Preprocess the key indicator information to obtain standard indicator information and store it in the data warehouse;

[0054] Step 3: Substitute the standard indicator information and corresponding indicator fields of the data warehouse into the weight model to obtain the first weights corresponding to different indicator fields;

[0055] Step 4: Use the historical bidding data in the data warehouse as the training set and the standard bidding documents as the test set. Use the training set to train the bid evaluation model. Use the test set and the first weights of different indicator fields to update the latest indicator parameters of the bid evaluation model to obtain the final bid evaluation model.

[0056] Step 5: Build the final bid evaluation model into the bid evaluation brain system to form the bid evaluation brain based on the big data model. Substitute the actual bidding documents into the bid evaluation brain system to obtain the bid evaluation report results.

[0057] In this embodiment, the bidding project refers to a plan issued by an organization or institution to the market to invite bids for the purpose of purchasing products or services.

[0058] In this embodiment, the original data is the initial information related to the bidding project, including unprocessed original documents or data such as project requirements, bidding documents, and supplier information.

[0059] In this embodiment, the preset field refers to a data field that is predefined during the data processing process and is used to extract and analyze key information.

[0060] In this embodiment, natural language processing technology is a technology in the field of computer science and artificial intelligence, which aims to enable computers to understand, interpret, generate and process human language. NLP combines methods from computational linguistics, computer science and statistics.

[0061] In this embodiment, the key indicator information refers to the core information of important data such as project requirements, bidding conditions, and bid evaluation criteria in the bidding documents.

[0062] In this embodiment, preprocessing is a processing method of converting raw data into a format more suitable for subsequent processing and analysis.

[0063] In this embodiment, the data warehouse is a system for centrally storing and managing data, and is mainly used for integrating data services.

[0064] In this embodiment, the indicator field refers to a variable in a data set used to describe and measure specific characteristics or performance. For example, in a bidding project, the indicator fields may include "bid amount", "company qualification", "completion time", etc.

[0065] In this embodiment, the weight model is a mathematical model used to assign importance to different indicators.

[0066] In this embodiment, the first weight represents the initial importance distribution of each indicator field in the bid evaluation process.

[0067] In this embodiment, the historical bidding data refers to relevant data collected from past bidding projects, including bidding plans, scoring results, bid evaluation criteria, etc.

[0068] In this embodiment, the training set refers to a historical bidding data set used to train the bid evaluation model.

[0069] In this embodiment, the standard bidding document refers to a reference document used to test the bid evaluation model, including a standardized bidding document.

[0070] In this embodiment, the test set refers to a set of standardized bidding documents with clear evaluation criteria, which are used to evaluate and verify the performance of the trained large bidding evaluation model.

[0071] In this embodiment, the bid evaluation model is a trained algorithm model that can automatically evaluate the quality of bid documents.

[0072] In this embodiment, the final bid evaluation model is a trained and optimized machine learning model that integrates historical bidding data and standard indicator information, and continuously updates parameters to accurately evaluate the quality of new bid documents.

[0073] In this embodiment, the bid evaluation brain system is a big data analysis system that integrates the final bid evaluation model and is used to automatically process and evaluate bidding documents.

[0074] In this embodiment, the big data model-based bid evaluation brain is an evaluation system that integrates big data analysis technology and machine learning models.

[0075] In this embodiment, the actual bidding document refers to the document of the actual bidding release, including project requirements, technical specifications, contract terms, bidding requirements, etc.

[0076] In this embodiment, the bid evaluation report result is an analysis and evaluation report generated by the bid evaluation brain system, which includes a detailed evaluation of each bid item in the actual bidding document.

[0077] The working principle and beneficial effects of the above technical solution are: through natural language processing and weight models, the bid evaluation process is automated and optimized to improve efficiency and accuracy. Data processing and model training enhance standardization and transparency, support data-driven decision-making, have strong adaptability, and significantly improve the quality of bid evaluation.

[0078] Embodiment 2:

[0079] Based on the above embodiment 1, original data related to the bidding project is collected, including:

[0080] Introduce blockchain technology, design blockchain network architecture, and preset node permissions and data sharing mechanisms of blockchain network;

[0081] Determine the main data source for bidding and develop the interface of the main data source according to the preset interface standards;

[0082] The original interface data of the bidding project is obtained through the blockchain subject market data interface, and the original data of the bidding project is obtained through data format conversion preprocessing operations, wherein the original data includes formatted data and unformatted data.

[0083] In this embodiment, blockchain technology is a distributed ledger technology that ensures the security and reliability of data through decentralization, encryption and consensus mechanisms.

[0084] In this embodiment, the blockchain network architecture includes: nodes, consensus mechanism, smart contracts, data structure, etc.

[0085] In this embodiment, node permissions determine the operational capabilities and access control of each node in the network.

[0086] In this embodiment, the data sharing mechanism determines how to transfer and access data between different nodes.

[0087] In this embodiment, the main data source refers to a system or service that provides bidding project data, such as bidding announcements, bidding records, contract information, etc.

[0088] In this embodiment, the main market data interface refers to an interface used to extract data from a main data source (such as a bidding platform).

[0089] In this embodiment, the original interface data is initial data obtained from the main data source through the interface.

[0090] In this embodiment, data format conversion is a process of converting data from one format to another format.

[0091] In this embodiment, formatted data refers to data that has been organized according to predetermined rules and structures, such as JSON, XML, CSV, etc.

[0092] In this embodiment, unformatted data refers to data without a specific structure or rule, such as free text, images, audio, etc.

[0093] The working principle and beneficial effects of the above technical solution are: data transparency and security are improved through node permissions and data sharing mechanisms. The main data source interfaces are developed in accordance with standards to ensure accurate acquisition of bidding data. Data processing includes formatting and conversion, improving data consistency and analysis capabilities, and overall improving system efficiency and trust.

[0094] Embodiment 3:

[0095] Based on the above embodiment 1, the key indicator information in the original data is identified and extracted using natural language processing technology according to the preset fields, including:

[0096] If the original data is formatted data, the corresponding field indicators in the original data are extracted using the field template and the field indicators are matched with the corresponding data results to generate the key indicator information of the field indicator-indicator data;

[0097] If the original data is unformatted data, the unformatted data is processed using an OCR tool, feature processing is performed on the unformatted data, useless data is filtered out, and a first feature sequence is obtained;

[0098] Match any unit feature of the first feature sequence with a standard dictionary, obtain the character with the highest matching degree as the unit character, and obtain text character data corresponding to the first feature sequence;

[0099] Using NER technology to identify entities of the text character data in combination with preset fields, and marking the entities according to the field type of the preset fields;

[0100] According to the annotation information of the entity, the entity is classified to obtain the key indicator information of the original data.

[0101] In this embodiment, the field template is a predefined structure or format for extracting specific information from the data.

[0102] In this embodiment, the OCR tool is a technical tool for extracting and recognizing text information from scanned documents, images or other unstructured data.

[0103] In this embodiment, feature processing is to select and filter the most useful features for matching the unformatted data and eliminate redundant or irrelevant features.

[0104] In this embodiment, the first feature sequence refers to a preliminary feature set extracted from the unformatted data processed by OCR.

[0105] In this embodiment, the unit feature is an individual feature item extracted from the first feature sequence, such as a single character, word or symbol.

[0106] In this embodiment, the standard dictionary is a reference database containing known characters, words or symbols and their definitions or codes.

[0107] In this embodiment, the matching degree refers to the similarity between a unit feature and an entry in the standard dictionary.

[0108] In this embodiment, the unit character refers to the smallest identification unit extracted from the data.

[0109] In this embodiment, the text character data refers to the processed text content extracted from the non-formatted data.

[0110] In this embodiment, the NER technology is a technology in natural language processing, which is used to identify specific types of entities from text, such as names of people, places, organizations, time, etc.

[0111] In this embodiment, an entity refers to a word or phrase in a text that has a specific meaning and represents a specific object, concept or information.

[0112] In this embodiment, labeling refers to assigning a label or category to a specific part of the data (such as text or an object in an image).

[0113] The working principle and beneficial effects of the above technical solution are: extracting formatted data indicators through field templates, using OCR processing and feature matching to convert non-formatted data, combining NER technology for entity recognition and labeling, and finally automatically classifying and generating key indicator information. This improves data accuracy, processing efficiency and automation level.

[0114] Embodiment 4:

[0115] On the basis of the above-mentioned embodiment 1, the key indicator information is preprocessed to obtain standard indicator information, which is then stored in the data warehouse, including:

[0116] The key indicator information is filtered to remove abnormal data, and then pre-processed according to the preset standard unit to obtain standard indicator information data;

[0117] Design the data table structure according to the preset standard fields and successfully match the data table with the data warehouse;

[0118] The standard indicator information data is stored in batches in the data table, and the data that does not conform to the data table fields is filtered into the junk table as useless data;

[0119] Complete the data storage of the data warehouse and start it using the SQL script.

[0120] In this embodiment, abnormal data refers to data that significantly deviates from a normal range in a data set.

[0121] In this embodiment, the preset standard unit is a predefined unit standard. For example, when processing financial data, the amount may be unified into "US dollars" or "Euros".

[0122] In this embodiment, the data table is a structured collection in a database, used to store data.

[0123] In this embodiment, the garbage table is a table temporarily or specifically used to store data that does not meet the preset standard.

[0124] In this embodiment, the SQL script is a set of instructions written in SQL for automating the database.

[0125] The working principle and beneficial effects of the above technical solution are: by screening abnormal data, storing standard indicator information in a designed data table after pre-processing, transferring useless data into a garbage table, and finally using SQL scripts to complete data storage, data quality and management efficiency are improved, ensuring data cleanliness and automated storage processes.

[0126] Embodiment 5:

[0127] On the basis of the above-mentioned embodiment 1, the standard indicator information of the data warehouse and the corresponding indicator fields are substituted into the weight model to obtain the first weights corresponding to different indicator fields, including:

[0128] Obtain the preset standard fields of the data warehouse and design the basic information of the scoring table; then combine any indicator field of the standard indicator information with the scoring criteria in the scoring table for evaluation to obtain the first score;

[0129] Where A1 represents the first score of the indicator field, j represents the jth scoring standard, m represents that the scoring table has m scoring standards in total, i represents the i-th indicator data value of the indicator field, n represents that the field has n indicator data values ​​in total, and L ij represents the logical length difference between the i-th indicator data value and the j-th scoring standard, σ1 represents the logical length difference between the i-th indicator data value and the j-th scoring standard, and σ2 represents the logical length difference between the i-th indicator data value and the j-th scoring standard. ij The variance of L j represents the logical length of the jth scoring criterion, f() represents the hierarchical superposition function, e represents the error coefficient, and w ij represents the hierarchical distance difference between the i-th indicator data value and the j-th scoring standard, w max Indicates the maximum length of the layer, w ave represents the maximum mean of the level;

[0130] Finally, the first scores of all indicators of the standard indicator information are obtained.

[0131] In this embodiment, the basic information includes: field name, field type, field constraint and other contents.

[0132] In this embodiment, the scoring criteria include information describing the logical length, hierarchical length, variance, error coefficient, etc. of the scoring criteria.

[0133] In this embodiment, the first score is a calculated value used to measure the score result of a certain indicator data value according to the scoring standard.

[0134] In this embodiment, the logical length refers to the logical or quantitative "distance" or "range" of a certain data value or scoring criterion.

[0135] In this embodiment, the hierarchy is because the scoring table contains multiple scoring criteria, and these criteria may be arranged in a certain logical order or priority to form a different hierarchy.

[0136] The working principle and beneficial effects of the above technical solution are: by designing a scoring table and combining standard indicator information with scoring criteria, the first score of each indicator is calculated. Factors such as logical length distance difference, variance and hierarchical distance difference are considered to obtain a comprehensive score, ensuring the accuracy, consistency and reliability of the bid evaluation results and improving the accuracy of data analysis.

[0137] Embodiment 6:

[0138] On the basis of the above-mentioned embodiment 5, obtaining the first weights corresponding to different indicator fields further includes:

[0139] Substitute the first scores of all indicator fields of the standard indicator information and their corresponding indicator fields into the weight model;

[0140] Wherein, g represents the first weight of the indicator field, n1 represents a total of n1 indicator fields, σ i1 represents the initial evaluation weight of the i1th indicator field, A i1 represents the first score of the i1th indicator field, and lg() represents the weight model;

[0141] The first weights corresponding to different indicator fields of the standard indicator information are obtained.

[0142] In this embodiment, the initial evaluation weight

[0143] The working principle and beneficial effect of the above technical solution are: the first score of the indicator field is substituted into the weight model to calculate the weight of each field, and the final weight is obtained by using the initial evaluation weight and the score value. This method optimizes the weight distribution, improves the accuracy of the scoring system, improves the versatility and practicality of the tool, and greatly improves the bidding efficiency.

[0144] Embodiment 7:

[0145] On the basis of the above-mentioned embodiment 1, the historical bidding data of the data warehouse is used as a training set, and the standard bidding documents are used as a test set. The training set is used to train the bid evaluation model, and the test set and the first weights of different indicator fields are used to update the latest indicator parameters of the bid evaluation model to obtain the final bid evaluation model, including:

[0146] Extract a data set containing bidding history records from the data warehouse as a training set, clean and filter the training set data, and then extract features from the training set to obtain different feature fields and their corresponding data;

[0147] Using the characteristic fields to perform a first training on the large bid evaluation model to obtain a first model indicator;

[0148] Afterwards, the large training and bid evaluation model is iteratively processed to obtain the model index corresponding to the number of iterations, and the model index and the first model index constitute the model index set;

[0149] The first standard parameter of the large bid evaluation model is obtained by using the model indicator set, and the standard bidding document is used as a test set for testing and evaluation to obtain the evaluation indicator of the large bid evaluation model;

[0150] According to the evaluation index and the first weights of different index fields, the first standard parameters of the large bid evaluation model are adjusted to obtain the hyper parameters of the large bid evaluation model;

[0151] After the hyperparameters are used to update the bid evaluation model, the final bid evaluation model is obtained.

[0152] In this embodiment, the feature field refers to a variable or attribute extracted from raw data and used to describe or characterize data characteristics.

[0153] In this embodiment, the model index is an important parameter metric used to evaluate the model performance.

[0154] In this embodiment, iteration refers to adjusting model parameters and updating the model multiple times to improve performance when training the model.

[0155] In this embodiment, the model indicator set refers to the summary of performance indicators of the model collected during the training and iteration process.

[0156] In this embodiment, the first standard parameter is a parameter determined based on performance indicators during initial training and iteration.

[0157] In this embodiment, the evaluation index refers to various metrics used to measure and evaluate the performance of the large bidding evaluation model.

[0158] In this embodiment, the hyperparameters are model training parameters, including: learning rate, batch size, regularization parameters, etc.

[0159] The working principle and beneficial effects of the above technical solution are: extract bidding records from the data warehouse, clean and extract features, and then train the bid evaluation model. Iterate the training to obtain the model indicator set, and use the evaluation indicators to adjust the standard parameters, optimize the hyperparameters, improve the model accuracy, and ensure its effectiveness in practical applications.

[0160] Embodiment 8:

[0161] On the basis of the above-mentioned embodiment 7, the final bid evaluation big model is built into the bid evaluation brain system to form a big data model-based bid evaluation brain. The actual bidding documents are substituted into the bid evaluation brain system to obtain the bid evaluation report results, including:

[0162] Build the final bid evaluation model into the bid evaluation brain system to form the bid evaluation brain based on the big data model;

[0163] Create an API interface in the bid evaluation brain system and connect it to the front-end interface; upload the actual bidding documents through the front-end interface, generate initial bid evaluation results through the final bid evaluation model processing, use the business integration system to integrate the initial bid evaluation results into business, obtain the bid evaluation report results and output them to the front-end, and save the initial bid evaluation results in the back-end database.

[0164] In this embodiment, the bid evaluation brain system is the most important analysis module in the intelligent evaluation system. Figure 2 shown.

[0165] In this embodiment, the API interface is a set of protocols and tools that define how software systems communicate with each other.

[0166] In this embodiment, the front-end interface is the visual part of the user's interaction with the system, including a web page or an application program interface.

[0167] In this embodiment, the initial bid evaluation results refer to the preliminary evaluation results generated by the final bid evaluation model during the bid evaluation process, including: scores, rankings, classifications, etc.

[0168] The working principle and beneficial effects of the above technical solution are: integrate the final bid evaluation model into the bid evaluation brain system, create an API interface to connect with the front end. The front end uploads the bidding documents, the model generates the initial bid evaluation results, outputs the bid evaluation report after business integration, and saves the data. Improve the bid evaluation efficiency and accuracy, optimize data processing and report generation, and realize the intelligent management of the entire bidding process

[0169] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent bid evaluation method based on big data, characterized in that: include: Step 1: Collect raw data related to the bidding project, and use natural language processing technology to identify and extract key indicator information from the raw data according to preset fields; Step 2: Preprocess the key indicator information to obtain standard indicator information and store it in the data warehouse; Step 3: Substitute the standard indicator information and corresponding indicator fields of the data warehouse into the weight model to obtain the first weights corresponding to different indicator fields; Step 4: Use the historical bidding data in the data warehouse as the training set and the standard bidding documents as the test set. Use the training set to train the bid evaluation model. Use the test set and the first weights of different indicator fields to update the latest indicator parameters of the bid evaluation model to obtain the final bid evaluation model. Step 5: Build the final bid evaluation model into the bid evaluation brain system to form the bid evaluation brain based on the big data model. Substitute the actual bidding documents into the bid evaluation brain system to obtain the bid evaluation report results.

2. According to the big data-based intelligent bidding evaluation method of claim 1, it is characterized in that: Collect original data related to the bidding project, including: Introduce blockchain technology, design blockchain network architecture, and preset node permissions and data sharing mechanisms of blockchain network; Determine the main data source for bidding and develop the interface of the main data source according to the preset interface standards; The original interface data of the bidding project is obtained through the blockchain subject market data interface, and the original data of the bidding project is obtained through data format conversion preprocessing operations, wherein the original data includes formatted data and unformatted data.

3. According to the intelligent bidding evaluation method based on big data according to claim 1, it is characterized in that: Use natural language processing technology to identify and extract key indicator information from raw data according to preset fields, including: If the original data is formatted data, the corresponding field indicators in the original data are extracted using the field template and the field indicators are matched with the corresponding data results to generate the key indicator information of the field indicator-indicator data; If the original data is unformatted data, the unformatted data is processed using an OCR tool, feature processing is performed on the unformatted data, useless data is filtered out, and a first feature sequence is obtained; Match any unit feature of the first feature sequence with a standard dictionary, obtain the character with the highest matching degree as the unit character, and obtain text character data corresponding to the first feature sequence; Using NER technology to identify entities of the text character data in combination with preset fields, and marking the entities according to the field type of the preset fields; According to the annotation information of the entity, the entity is classified to obtain the key indicator information of the original data.

4. According to the big data-based intelligent bidding evaluation method of claim 1, it is characterized in that: After preprocessing the key indicator information, standard indicator information is obtained and stored in the data warehouse, including: The key indicator information is filtered to remove abnormal data, and then pre-processed according to the preset standard unit to obtain standard indicator information data; Design the data table structure according to the preset standard fields and successfully match the data table with the data warehouse; The standard indicator information data is stored in batches in the data table, and the data that does not conform to the data table fields is filtered into the junk table as useless data; Complete the data storage of the data warehouse and start it using the SQL script.

5. According to the intelligent bidding evaluation method based on big data according to claim 1, it is characterized in that: Substitute the standard indicator information and corresponding indicator fields of the data warehouse into the weight model to obtain the first weights corresponding to different indicator fields, including: Obtain the preset standard fields of the data warehouse and design the basic information of the scoring table; then combine any indicator field of the standard indicator information with the scoring criteria in the scoring table for evaluation to obtain the first score; Where A1 represents the first score of the indicator field, j represents the jth scoring standard, m represents that the scoring table has m scoring standards in total, i represents the i-th indicator data value of the indicator field, n represents that the field has n indicator data values ​​in total, and L ij represents the logical length difference between the i-th indicator data value and the j-th scoring standard, σ1 represents the logical length difference between the i-th indicator data value and the j-th scoring standard, and σ2 represents the logical length difference between the i-th indicator data value and the j-th scoring standard. ij The variance of L j represents the logical length of the jth scoring criterion, f() represents the hierarchical superposition function, e represents the error coefficient, and w ij represents the hierarchical distance difference between the i-th indicator data value and the j-th scoring standard, w max Indicates the maximum length of the layer, w ave represents the maximum mean of the level; Finally, the first scores of all indicators of the standard indicator information are obtained.

6. According to the big data-based intelligent bidding evaluation method of claim 5, it is characterized in that: Obtaining the first weights corresponding to different indicator fields also includes: Substitute the first scores of all indicator fields of the standard indicator information and their corresponding indicator fields into the weight model; Wherein, g represents the first weight of the indicator field, n1 represents a total of n1 indicator fields, σ i1 represents the initial evaluation weight of the i1th indicator field, A i1 represents the first score of the i1th indicator field, and 1g() represents the weight model; The first weights corresponding to different indicator fields of the standard indicator information are obtained.

7. The intelligent bid evaluation method based on big data according to claim 1 is characterized in that: The historical bidding data in the data warehouse is used as the training set, and the standard bidding documents are used as the test set. The training set is used to train the bid evaluation model. The test set and the first weights of different indicator fields are used to update the latest indicator parameters of the bid evaluation model to obtain the final bid evaluation model, including: Extract a data set containing bidding history records from the data warehouse as a training set, clean and filter the training set data, and then extract features from the training set to obtain different feature fields and their corresponding data; Using the characteristic fields to perform a first training on the large bid evaluation model to obtain a first model indicator; Afterwards, the large training and bid evaluation model is iteratively processed to obtain the model index corresponding to the number of iterations, and the model index and the first model index constitute a model index set; The first standard parameter of the large bid evaluation model is obtained by using the model indicator set, and the standard bidding document is used as a test set for testing and evaluation to obtain the evaluation indicator of the large bid evaluation model; According to the evaluation index and the first weights of different index fields, the first standard parameters of the large bid evaluation model are adjusted to obtain the hyper parameters of the large bid evaluation model; After the hyperparameters are used to update the bid evaluation model, the final bid evaluation model is obtained.

8. The intelligent bid evaluation method based on big data according to claim 7 is characterized in that: The final bid evaluation model is built into the bid evaluation brain system to form the bid evaluation brain based on the big data model. The actual bidding documents are substituted into the bid evaluation brain system to obtain the bid evaluation report results, including: Build the final bid evaluation model into the bid evaluation brain system to form the bid evaluation brain based on the big data model; Create an API interface in the bid evaluation brain system and connect it to the front-end interface; upload the actual bidding documents through the front-end interface, generate initial bid evaluation results through the final bid evaluation model processing, use the business integration system to integrate the initial bid evaluation results into business, obtain the bid evaluation report results and output them to the front-end, and save the initial bid evaluation results in the back-end database.

Images