Performance evaluation method, system, equipment and medium based on procurement major large model

Abstract

The invention discloses an efficiency evaluation method, system, equipment and medium based on a procurement major large model, and relates to the technical field of efficiency evaluation. The performance evaluation method comprises the steps of obtaining data of a to-be-evaluated project; preprocessing the data of the to-be-assessed project to obtain preprocessed to-be-assessed data; determining an evaluation model according to the to-be-evaluated data and a preset procurement major large model; and performing efficiency evaluation on the to-be-evaluated project according to the evaluation model and the procurement major large model. According to the method, the evaluation report can be quickly generated, and the optimization suggestion can be given, so that a worker can perform targeted improvement according to the evaluation report, continuous optimization of the enterprise energy efficiency model is realized, the efficiency of purchasing in production and operation can be better and continuously exerted, and the working efficiency of the worker is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of performance evaluation, and in particular to a performance evaluation method based on a large model of procurement professionals, a performance evaluation system based on a large model of procurement professionals, a computer device and a readable storage medium. Background Art

[0002] At present, the methods of purchasing project evaluation mainly focus on the transaction price to determine the purchasing effect, the purchasing time to determine the purchasing efficiency, and the performance evaluation to determine the purchasing quality. Although they can reflect the purchasing effect from their respective perspectives to a certain extent, they have not risen to the level of correctly analyzing and understanding the "effectiveness" and "capability" of purchasing projects, making it difficult to better and continuously exert the effectiveness of purchasing in production and operation. Summary of the invention

[0003] The technical problem to be solved by the present invention is that the current procurement project evaluation method is difficult to better and continuously exert the effectiveness of procurement in production and operation.

[0004] In view of the above-mentioned deficiencies of the prior art, the following solutions are provided:

[0005] In a first aspect, the present invention provides a performance evaluation method based on a procurement professional big model, comprising: obtaining data of a project to be evaluated. Preprocessing the data of the project to be evaluated to obtain the preprocessed data to be evaluated. Determining an evaluation model based on the data to be evaluated and a preset procurement professional big model. And, performing performance evaluation on the project to be evaluated based on the evaluation model and the procurement professional big model. The data of the project to be evaluated includes a purchase order, a procurement project, procurement implementation, procurement results, bidding information, contract information, and approval information of the project to be evaluated.

[0006] Optionally, the data of the project to be evaluated is preprocessed to obtain the preprocessed data to be evaluated, including: using a transformer model (Bidirectional and Auto-Regressive Transformers, BART) model with information denoising and bidirectional language modeling and autoregressive mechanism to perform denoising, structuring and cleaning on the data of the project to be evaluated to obtain the preliminarily processed data. And, the preliminarily processed data is automatically labeled by a bidirectional encoder representation model from transformer (BERT) BERT-base model to obtain the automatically labeled data, and the preprocessed data to be evaluated includes the automatically labeled data.

[0007] Optionally, preprocessing the data of the project to be evaluated to obtain preprocessed data to be evaluated also includes: manually verifying the automatically marked data to obtain manually verified marked data, and the preprocessed data to be evaluated includes the manually verified marked data.

[0008] Optionally, the preliminarily processed data is automatically labeled using a BERT-base model, including: assigning one or more intent labels to the preliminarily processed data according to an intent recognition algorithm.

[0009] Optionally, determining the evaluation model according to the data to be evaluated and the preset procurement professional model includes: inputting the data to be evaluated into the procurement professional model. Identifying the correlation between each dimension of the multiple procurement evaluation dimensions in the procurement professional model and the data to be evaluated by calculating the Pearson correlation coefficient, and determining each dimension of the evaluation model according to the identification correlation result. Analyzing the influence of each of the multiple influencing factors under the multiple dimensions in the procurement professional model on the procurement results of the project to be evaluated by a nonlinear regression analysis algorithm and the data to be evaluated, and determining at least one influencing factor under each dimension of the evaluation model according to the analysis result. And, determining the evaluation model according to each dimension of the evaluation model and at least one influencing factor under each dimension of the evaluation model. Wherein, each dimension of the evaluation model includes at least one procurement evaluation dimension among each dimension of the multiple procurement evaluation dimensions in the procurement professional model whose correlation with the project to be evaluated is higher than a first preset threshold. At least one influencing factor under each dimension of the evaluation model is at least one influencing factor among the multiple influencing factors under the multiple dimensions of the multiple procurement evaluation dimensions in the procurement professional model whose influence on the procurement results of the project to be evaluated is greater than a second preset threshold.

[0010] Optionally, determining the evaluation model according to each dimension of the evaluation model and at least one influencing factor under each dimension of the evaluation model includes: determining the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model. And determining the evaluation model according to each dimension of the evaluation model, at least one influencing factor under each dimension of the evaluation model, and the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model.

[0011] Optionally, determining the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model includes: using Fourier transform method, random forest regression method, and spatiotemporal joint model method to construct weight evaluation models for the evaluation model, and obtaining three weight evaluation models. The three weight evaluation models are verified according to the verification data set to obtain verification results. And, determining the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model according to the verification results. Among them, the verification data set is obtained by manual verification based on the preprocessed data to be evaluated.

[0012] Optionally, the three weight evaluation models are verified respectively according to the verification data set to obtain verification results, including: obtaining the baseline performance F1 value of each influencing factor under each dimension of the three weight evaluation models respectively according to the verification data set. And, randomly arrange each influencing factor under each dimension of the three weight evaluation models respectively, and record the performance F1 value of each influencing factor under each dimension of the three weight evaluation models after scrambling. Among them, the verification results include the baseline performance F1 value of each influencing factor under each dimension of the three weight evaluation models and the performance F1 value of each influencing factor under each dimension of the three weight evaluation models after scrambling.

[0013] Optionally, the weight proportion of each influencing factor in at least one influencing factor under each dimension of the evaluation model is determined according to the verification result, including: comparing the performance F1 value of each influencing factor under each dimension of the three weight evaluation models after the scrambling with the benchmark performance F1 value of each influencing factor under each dimension of the three weight evaluation models, and obtaining the difference results of each influencing factor under each dimension of the three weight evaluation models. And, according to the difference results of each influencing factor under each dimension of the three weight evaluation models, the weight proportion of each influencing factor in at least one influencing factor under each dimension of the evaluation model is determined.

[0014] Optionally, a performance evaluation is performed on the project to be evaluated based on the evaluation model and the procurement professional big model, including: inputting at least one influencing factor under each dimension of the evaluation model, data corresponding to each influencing factor in the project to be evaluated and at least one influencing factor under each dimension of the evaluation model, and the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model into the procurement professional big model, so that the procurement professional big model can summarize the project to be evaluated by using text comprehension and text generation capabilities, and output performance evaluation results, improvement measures and optimization strategies.

[0015] Optionally, the method further includes: inputting at least one influencing factor under each dimension of the evaluation model, data corresponding to each influencing factor in at least one influencing factor under each dimension of the evaluation model in similar projects, and the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model into the procurement professional big model, optimizing the evaluation model according to the project results of similar projects and the output results of the procurement professional big model, and obtaining an optimized evaluation model. Wherein, similar projects are historical projects similar to the project to be evaluated. Evaluating the performance of the project to be evaluated according to the evaluation model and the procurement professional big model includes: evaluating the performance of the project to be evaluated according to the optimized evaluation project and the procurement professional big model.

[0016] Optionally, the method for acquiring the procurement professional big model includes: acquiring raw data. Structuring the raw data to obtain data after structured processing. Performing conversion preprocessing on the data after structured processing to obtain data after conversion preprocessing. Fine-tuning the parameters of the preset big model to obtain a fine-tuned big model. And, using the data after conversion preprocessing to perform model training on the fine-tuned big model to obtain the procurement professional big model. The raw data includes procurement professional knowledge documents and data of historical procurement projects.

[0017] Optionally, the data after structured processing is converted and preprocessed, including: using embedding technology to extract the meaning of each sentence in the data after structured processing, calculating the correlation between each sentence, obtaining multiple data blocks according to the correlation between each sentence, performing knowledge slicing on each data block in the multiple data blocks to obtain sliced ​​data, using the term frequency-inverse document frequency TF-IDF algorithm to extract keywords from the sliced ​​data, and performing vectorization conversion and vector entry on the sliced ​​data to obtain the data after conversion preprocessing.

[0018] In a second aspect, the present invention provides a performance evaluation system based on a procurement professional big model, comprising: a data acquisition module for the project to be evaluated, a preprocessing module, an evaluation model determination module and a performance evaluation module. Among them, the data acquisition module for the project to be evaluated is configured to: obtain the data of the project to be evaluated, and the data of the project to be evaluated includes the purchase order, procurement project, procurement implementation, procurement results, bidding information, contract information and approval information of the project to be evaluated. The preprocessing module is configured to: preprocess the data of the project to be evaluated to obtain the preprocessed data to be evaluated. The evaluation model determination module is configured to: determine the evaluation model according to the data to be evaluated and the preset procurement professional big model. The performance evaluation module is configured to: perform performance evaluation on the project to be evaluated according to the evaluation model and the procurement professional big model.

[0019] In a third aspect, the present invention provides a computer device including a memory and a processor, wherein a computer program is stored in the memory. When the processor runs the computer program stored in the memory, the processor executes the above-mentioned performance evaluation method based on the procurement professional big model.

[0020] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the processor executes the above-mentioned performance evaluation method based on the procurement professional big model.

[0021] Through the performance evaluation method, system, equipment and medium based on the procurement professional large model provided by the present invention, an evaluation model can be dynamically and adaptively constructed according to the project to be evaluated, thereby improving the scope of application of the present invention. Provide multiple evaluation dimensions, and based on the results of historical data analysis, intelligently determine the weights between different indicators under each dimension and the weights of each dimension, generate a general model and a benchmark model, so that users can select an evaluation model according to actual conditions. The present invention can quickly generate an evaluation report and give optimization suggestions so that staff can make targeted improvements based on the evaluation report, realize continuous optimization of the enterprise's energy efficiency model, thereby better and continuously exerting the effectiveness of procurement in production and operation, and improving personnel work efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a flow chart of a method for evaluating the effectiveness of a procurement professional model in an embodiment of the present invention;

[0023] Figure 2 It is a flowchart of another performance evaluation method based on a large procurement professional model in an embodiment of the present invention;

[0024] Figure 3 It is a flowchart of another method for evaluating the effectiveness of a procurement professional model in an embodiment of the present invention;

[0025] Figure 4 It is a flowchart of another performance evaluation method based on a large procurement professional model in an embodiment of the present invention;

[0026] Figure 5 It is a flowchart of another method for evaluating the effectiveness of a procurement professional model in an embodiment of the present invention;

[0027] Figure 6 It is a flowchart of another performance evaluation method based on a large procurement professional model in an embodiment of the present invention;

[0028] Figure 7 It is a flowchart of another method for evaluating the effectiveness of a procurement professional model in an embodiment of the present invention;

[0029] Figure 8It is a flowchart of another performance evaluation method based on a large procurement professional model in an embodiment of the present invention;

[0030] Fig. 9 It is an architecture diagram of a performance evaluation system based on a large procurement professional model in an embodiment of the present invention;

[0031] Fig.10 It is an architecture diagram of another performance evaluation system based on a large procurement professional model in an embodiment of the present invention;

[0032] Fig.11 It is an architecture diagram of a performance evaluation system based on a large procurement professional model in an example of an embodiment of the present invention;

[0033] Fig.12 It is an architecture diagram of another performance evaluation system based on a large procurement professional model in an example of an embodiment of the present invention;

[0034] Fig.13 The figure is a structural diagram of a computer device in an embodiment of the present invention. DETAILED DESCRIPTION

[0035] In order to enable those skilled in the art to better understand the technical solution of the present invention, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0036] It should be understood that the specific embodiments and drawings described herein are only used to explain the present invention rather than to limit the present invention.

[0037] It can be understood that, in the absence of conflict, the various embodiments of the present invention and the various features in the embodiments can be combined with each other.

[0038] It can be understood that, for the convenience of description, the drawings of the present invention only show the parts related to the present invention, while the parts irrelevant to the present invention are not shown in the drawings.

[0039] It can be understood that each unit and module involved in the embodiments of the present invention may correspond to only one physical structure, or may be composed of multiple physical structures, or multiple units and modules may be integrated into one physical structure.

[0040] It can be understood that, without conflict, the functions and steps marked in the flowcharts and block diagrams of the present invention may occur in an order different from that marked in the drawings.

[0041] It is understood that the flowcharts and block diagrams of the present invention illustrate the possible architectures, functions, and operations of the systems, devices, equipment, and methods according to the various embodiments of the present invention. Each box in the flowchart or block diagram may represent a unit, module, program segment, or code, which contains executable instructions for implementing the specified functions. Moreover, each box or combination of boxes in the block diagram and flowchart may be implemented by a hardware-based system that implements the specified functions, or may be implemented by a combination of hardware and computer instructions.

[0042] It can be understood that the units and modules involved in the embodiments of the present invention can be implemented by software or hardware. For example, the units and modules can be located in a processor.

[0043] The embodiment of the present invention provides a method for evaluating the effectiveness of a procurement professional model. Figure 1 As shown, the method includes steps 101 to 104.

[0044] Step 101: Obtain data of the project to be evaluated.

[0045] In step 101, the data of the project to be evaluated includes the purchase order, purchase project, purchase implementation, purchase result, bidding information, contract information and approval information of the project to be evaluated.

[0046] Step 102: preprocess the data of the project to be evaluated to obtain preprocessed data to be evaluated.

[0047] In some embodiments, Figure 2 As shown, the implementation method of step 102 includes step 1021 and step 1022.

[0048] Step 1021: Use the information denoising BART model to perform denoising, structuring and cleaning on the data of the project to be evaluated to obtain preliminarily processed data.

[0049] Understandably, the BART model uses a standard transformer-based sequence-to-sequence structure, the main difference being the use of Gaussian error linear unit GeLU activation function and normal distribution initialization parameters. BART combines a bidirectional Transformer encoder and a unidirectional autoregressive Transformer decoder, and is a typical denoising autoencoder that performs pre-training by denoising and reconstructing noisy input text. The pre-training process includes the introduction of a variety of noises, such as word masking, word deletion, sentence permutation transformation, document rotation transformation, and text filling. The model uses these noise training to improve its ability to recover the original text.

[0050] Step 1022: Automatically label the preliminarily processed data using the BERT-base model to obtain automatically labeled data.

[0051] BERT (Bi Direction Encoder Representations from Transformers) is a language model architecture based on Transformer. It consists of multiple layers of self-attention and feedforward neural networks. BERT uses a bidirectional approach to capture contextual information from the previous and next words in a sentence. Depending on the scale of the model architecture, BERT has four pre-trained versions, and the BERT-Base model is one of them.

[0052] According to step 1022, the pre-processed data to be evaluated includes automatically labeled data. The pre-processed data to be evaluated includes manually verified labeled data.

[0053] In some embodiments, the implementation method of step 1022 includes: assigning one or more intent tags to the initially processed data according to an intent recognition algorithm.

[0054] It can be understood that assigning one or more intent labels to the data after preliminary processing can ensure the accuracy and comprehensiveness of the labels.

[0055] In some embodiments, the implementation method of step 102 further includes: manually verifying the automatically labeled data to obtain manually verified labeled data.

[0056] Understandably, manual verification is equivalent to adding human experience, which can make the data labeling more accurate.

[0057] Step 103: Determine the evaluation model based on the data to be evaluated and the preset procurement professional model.

[0058] In some embodiments, Figure 3 As shown, the implementation method of step 103 includes steps 1031 to 1034.

[0059] Step 1031: Input the data to be evaluated into the procurement professional model.

[0060] Step 1032: Identify the correlation between each dimension of multiple procurement evaluation dimensions in the procurement professional model and the data to be evaluated by calculating the Pearson correlation coefficient, and determine each dimension of the evaluation model based on the identified correlation results.

[0061] In step 1032, each dimension of the evaluation model is at least one procurement evaluation dimension among the multiple procurement evaluation dimensions in the procurement professional large model, the correlation with the project to be evaluated being higher than a first preset threshold.

[0062] For example, a large procurement model can store multiple procurement evaluation dimensions. For example, there are five procurement evaluation dimensions: procurement timeliness, competitive effectiveness, economic benefits, contract performance assurance, and procurement compliance. For different projects to be evaluated, one or more dimensions can be determined from these five dimensions. The evaluation dimensions of the evaluation model for each project to be evaluated can be different, that is, the evaluation model can be customized, so that customized evaluation models can be provided for different procurement projects.

[0063] Understandably, procurement timeliness refers to the requirements and regulations on supplier response time, delivery time, product quality, etc. during the procurement process. In procurement management, procurement timeliness is an important indicator that can affect the production and operation of an enterprise. Therefore, when formulating procurement plans and procurement contracts, enterprises should fully consider procurement timeliness and effectively control and manage it.

[0064] It can be understood that competition effectiveness refers to the fact that through the competition mechanism in the procurement process, the purchaser can obtain better procurement results and benefits. In procurement activities, by introducing competition, suppliers can be encouraged to provide better products and services and reduce procurement costs. Introducing a competition mechanism in the procurement process can achieve better procurement results and benefits.

[0065] It can be understood that economic efficiency means that the purchaser can obtain the maximum benefit at the lowest cost through reasonable methods and means in the procurement process. Introducing competition mechanisms and reasonable negotiations in the procurement process can achieve better economic benefits.

[0066] Understandably, performance assurance mainly refers to the evaluation and assurance of the supplier's ability to perform the contract during the procurement process to ensure that the supplier can perform its obligations in accordance with the contract. During the procurement process, it is necessary to evaluate and ensure the supplier's performance ability to ensure that the supplier can perform its obligations in accordance with the contract, thereby protecting the company's economic and production benefits.

[0067] Understandably, procurement compliance mainly refers to the need for procurement personnel to comply with established procurement rules, procedures and policies during the procurement process. Procurement compliance is an important guarantee to ensure that corporate procurement activities are legal, compliant and transparent, and is also a necessary condition for companies to achieve sustainable development. Improving the compliance of procurement activities can reduce the legal and ethical risks of enterprises.

[0068] It can be understood that the first preset threshold can be set manually.

[0069] Exemplarily, the relevance of each dimension of the procurement evaluation dimension to the project to be evaluated may also be manually evaluated, and each dimension of the evaluation model may be determined.

[0070] Step 1033: Analyze the influence of multiple influencing factors in multiple dimensions of multiple procurement evaluation dimensions in the procurement professional large model on the procurement results of the project to be evaluated through a nonlinear regression analysis algorithm and the data to be evaluated, and determine at least one influencing factor in each dimension of the evaluation model based on the analysis results.

[0071] In step 1033, at least one influencing factor under each dimension of the evaluation model includes at least one influencing factor among multiple influencing factors under each dimension of multiple procurement evaluation dimensions in the procurement professional model, whose influence on the procurement result of the project to be evaluated is greater than a second preset threshold.

[0072] Exemplarily, multiple procurement evaluation dimensions in the procurement professional model may include multiple influencing factors under each dimension. For example, the influencing factors under the procurement timeliness dimension include: procurement preparation, procurement implementation, contract signing, successful recruitment in one go, etc. The influencing factors under the competition effectiveness dimension include: the number of applicants / number of companies participating in the review, the number of companies that passed the review, technical and commercial competition, price competition, etc. The influencing factors under the economic efficiency dimension include: capital surplus, low price for close technology, low price for superior technology, and rejection of the lowest price. The influencing factors under the performance guarantee dimension include: post-supplier evaluation, on-time delivery, contractually agreed arrival time, arrival time after the supplier places an order, etc. The influencing factors under the procurement compliance dimension include: electronic bid evaluation room, subjective review weight, announcement time, and expert extraction, etc.

[0073] It can be understood that the second preset threshold can be set manually.

[0074] Exemplarily, the influence of each influencing factor under each dimension of the evaluation model on the procurement result of the project to be evaluated can also be manually evaluated, and at least one influencing factor under each dimension of the evaluation model can be determined.

[0075] It is understandable that step 1032 and step 1033 can be performed by purchasing a professional large model.

[0076] Step 1034: determine the evaluation model according to each dimension of the evaluation model and at least one influencing factor under each dimension of the evaluation model.

[0077] In some embodiments, Figure 4 As shown, the implementation method of step 1034 includes step 1035 and step 1036.

[0078] Step 1035: Determine the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model.

[0079] In some embodiments, Figure 5 As shown, the implementation method of step 1035 includes steps 501 to 503.

[0080] Step 501: Use the Fourier transform method, the random forest regression method, and the spatiotemporal joint model method to construct weight evaluation models for the evaluation model, and obtain three weight evaluation models.

[0081] Step 502: Verify the three weight evaluation models respectively according to the verification data set to obtain verification results.

[0082] In step 502, the verification data set is obtained by manually verifying the preprocessed data to be evaluated.

[0083] In some embodiments, Figure 6 As shown, the implementation method of step 502 includes steps 5021 and 5022.

[0084] Step 5021: Obtain the baseline performance F1 value of each influencing factor in each dimension of the three weight evaluation models according to the verification data set.

[0085] Step 5022: Randomly arrange the influencing factors under each dimension of the three weight evaluation models respectively, and record the performance F1 value of each influencing factor under each dimension of the three weight evaluation models after the shuffle.

[0086] In step 502, the verification result includes the baseline performance F1 value of each influencing factor under each dimension of the three weight evaluation models and the shuffled performance F1 value of each influencing factor under each dimension of the three weight evaluation models.

[0087] Step 503: Determine the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model according to the verification result.

[0088] In some embodiments, Figure 7 As shown, the implementation method of step 503 includes steps 5031 and 5032.

[0089] Step 5031:

[0090] The performance F1 values ​​of each influencing factor under each dimension of the three shuffled weight evaluation models and the benchmark performance F1 values ​​of each influencing factor under each dimension of the three weight evaluation models are compared respectively to obtain the difference results of each influencing factor under each dimension of the three weight evaluation models.

[0091] Step 5032: Determine the weight proportion of each influencing factor in at least one influencing factor under each dimension of the evaluation model according to the difference results of each influencing factor under each dimension of the three weight evaluation models.

[0092] It can be understood that for any of the influencing factors in each dimension of the three weight evaluation models, the greater the difference between the importance score and the benchmark performance F1 value, the greater the impact of the influencing factor on the model performance, and therefore the higher its importance. Therefore, the weight proportion of the influencing factor with a large difference can be increased according to the difference results.

[0093] In some embodiments, the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model can also be manually adjusted, for example, the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model can be adjusted according to expert experience.

[0094] Step 1036: determine the evaluation model according to each dimension of the evaluation model, at least one influencing factor under each dimension of the evaluation model, and the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model.

[0095] It can be understood that the evaluation model can be formed when each dimension of the evaluation model, at least one influencing factor under each dimension of the evaluation model, and the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model are determined.

[0096] Step 104: Perform performance evaluation on the project to be evaluated based on the evaluation model and the procurement professional model.

[0097] In some embodiments, the implementation method of step 104 includes: inputting at least one influencing factor under each dimension of the project model, data corresponding to each influencing factor in the project to be evaluated and at least one influencing factor under each dimension of the evaluation model, and the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model into the procurement professional big model, so that the procurement professional big model can summarize the projects to be evaluated by using text comprehension and text generation capabilities, and output performance evaluation results, improvement measures and optimization strategies.

[0098] It can be understood that the basic information of the project to be evaluated and at least one influencing factor under each dimension of the project model, the data corresponding to each influencing factor in the project to be evaluated and at least one influencing factor under each dimension of the evaluation model, and the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model can also be input into the procurement professional model at the same time, so that the final performance evaluation report can display the basic information of the project to be evaluated.

[0099] In some cases, the evaluation model generated according to the above steps 1031 to 1036 is a preliminary model, which is not necessarily the best model and may need further optimization.

[0100] In some embodiments, before step 104, the method further includes: inputting at least one influencing factor under each dimension of the evaluation model, data corresponding to each influencing factor in at least one influencing factor under each dimension of the evaluation model in similar projects, and the weight proportion of each influencing factor in at least one influencing factor under each dimension of the evaluation model into the procurement professional model, optimizing the evaluation model according to the project results of the similar projects and the output results of the procurement professional model, and obtaining an optimized evaluation model. Wherein, similar projects are historical projects similar to the project to be evaluated.

[0101] In this case, the implementation method of step 104 may be: performing performance evaluation on the project to be evaluated based on the optimized evaluation project and the procurement professional model.

[0102] It can be understood that the optimization of the evaluation model may be to adjust the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model.

[0103] It can be understood that similar projects are completed historical projects, which are complete from project data to project results, so similar projects can be used to optimize the evaluation model until the evaluation model and the results of similar projects have a high degree of match (it can be manually determined whether the two have a high degree of match). The number of similar projects can be one or more, that is, multiple similar projects can be used to optimize the evaluation model, and one of the similar projects can be used to verify the optimized evaluation model. If the optimized evaluation model can achieve a high degree of match with the results of most similar projects, the optimized evaluation model can be used to evaluate the project to be evaluated this time.

[0104] In some embodiments, Figure 8 As shown, the method for purchasing a professional large model includes steps 801 to 805.

[0105] Step 801: Obtain original data.

[0106] In step 801, the original data includes procurement expertise documents and data of historical procurement projects.

[0107] Step 802: Perform structured processing on the original data to obtain structured processed data.

[0108] Step 803: Perform conversion preprocessing on the data after the structured processing to obtain the data after the conversion preprocessing.

[0109] In some embodiments, the implementation method of step 803 includes: using embedding technology to extract the meaning of each sentence in the data after structured processing, calculating the correlation between each sentence, obtaining multiple data blocks according to the correlation between each sentence, performing knowledge slicing on each data block in the multiple data blocks to obtain sliced ​​data, using the term frequency-inverse document frequency TF-IDF algorithm to extract keywords from the sliced ​​data, and performing vectorization conversion and vector entry on the sliced ​​data to obtain the data after conversion preprocessing.

[0110] It can be understood that the conversion preprocessing operation can convert the text data in the original data into a sequence annotation token sequence, and convert the image data in the original data into a pixel matrix, that is, the data after the conversion preprocessing includes a token sequence and a pixel matrix.

[0111] Step 804: fine-tune the parameters of the preset large model to obtain a fine-tuned large model.

[0112] For example, the big model can use any open source big model, such as the Yuanjing big model provided by China Unicom Group. The parameters of the big model such as learning rate, batch size, training rounds, regularization term, etc. can be fine-tuned. For example, the prefix tuning method is used to construct a task-related virtual sequence virtualtokens as the prefix before the input token, and then only the parameters of the prefix part are updated during training.

[0113] Step 805: Use the converted preprocessed data to perform model training on the fine-tuned large model to obtain a procurement professional large model.

[0114] It is understandable that during the training process, performance indicators such as the accuracy and loss function of the large model can be regularly monitored to evaluate the training effect of the large model. The trained model can be evaluated using the procurement professional test set (which can be obtained based on the data preprocessed by the conversion in step 803) to verify the performance of the procurement professional large model. The performance of the model in terms of fitting, underfitting, and poor generalization can be analyzed manually based on the evaluation results, and the fine-tuning strategy of the model can be adjusted based on the evaluation results. The process of model training, evaluation, and optimization is repeated until the performance of the procurement professional large model is satisfactory. After obtaining the procurement professional large model, the trained procurement professional large model can be encapsulated into an API interface for user use.

[0115] The following is a specific example to illustrate an efficiency evaluation method based on a large procurement professional model provided by an embodiment of the present solution.

[0116] This example uses the procurement department's bidding project as an example to demonstrate the enterprise performance evaluation process. The enterprise only needs to configure once, and the report file can be automatically generated every month, which provides great convenience for the enterprise.

[0117] 1. Evaluation model creation

[0118] To create a performance evaluation model (i.e., evaluation model), you can choose to create a new model, copy an existing model, a general model, a benchmark model, or any other model. If you choose to create a new model, you need to customize the model variables and factors according to the five dimensions in the model center, and set the weights of each dimension and each factor; if you choose to copy an existing model, you can directly use the created model; if you choose to choose a general model, you can directly use the dimensions and factors intelligently generated by the platform based on data analysis, and intelligently assign weights; if you choose a benchmark model, you can use the high-quality model that currently has excellent application effects on the platform.

[0119] 2. Create a dataset

[0120] Determine the data sources involved in the model and the data processing scope, automatically create data sets in the system and perform data cleaning, data processing, and data output.

[0121] 3. Evaluation template selection

[0122] Select any performance report template provided in the platform and bind it to the evaluation model created in step 1.

[0123] 4. Generate performance report

[0124] After completing the report and model binding, the platform automatically adapts the data model and automatically generates analysis charts, etc., to generate a performance evaluation analysis report.

[0125] The embodiment of the present invention provides a performance evaluation system based on a large model of procurement professionals, such as Fig. 9 As shown, the performance evaluation system 900 includes a data acquisition module 901 for the project to be evaluated, a preprocessing module 902 , an evaluation model determination module 903 and a performance evaluation module 904 .

[0126] The data acquisition module 901 of the project to be evaluated is configured to: acquire the data of the project to be evaluated, the data of the project to be evaluated including the purchase order, purchase project, purchase implementation, purchase result, bidding information, contract information and approval information of the project to be evaluated.

[0127] The preprocessing module 902 is configured to preprocess the data of the item to be evaluated to obtain the preprocessed data to be evaluated.

[0128] The evaluation model determination module 903 is configured to determine the evaluation model based on the data to be evaluated and the preset procurement professional model.

[0129] The performance evaluation module 904 is configured to perform performance evaluation on the item to be evaluated according to the evaluation model.

[0130] In some embodiments, the preprocessing module 902 is configured to: use the information noise reduction BART model to perform noise reduction, structural processing and cleaning processing on the data of the project to be evaluated to obtain the preliminarily processed data. And, automatically label the preliminarily processed data through the BERT-base model to obtain the automatically labeled data, and the preprocessed data to be evaluated includes the automatically labeled data.

[0131] In some embodiments, the preprocessing module 902 is further configured to: manually verify the automatically labeled data to obtain manually verified labeled data, and the preprocessed data to be evaluated includes the manually verified labeled data.

[0132] In some embodiments, when the preliminarily processed data is automatically labeled by the BERT-base model, the preprocessing module 902 is configured to assign one or more intent labels to the preliminarily processed data according to the intent recognition algorithm.

[0133] In some embodiments, the evaluation model determination module 903 is configured to: input the data to be evaluated into the procurement professional model. The correlation between each dimension of the multiple procurement evaluation dimensions in the procurement professional model and the data to be evaluated is identified by calculating the Pearson correlation coefficient, and each dimension of the evaluation model is determined according to the identification correlation result. The influence degree of each of the multiple influencing factors under each dimension of the multiple procurement evaluation dimensions in the procurement professional model on the procurement result of the project to be evaluated is analyzed by a nonlinear regression analysis algorithm and the data to be evaluated, and at least one influencing factor under each dimension of the evaluation model is determined according to the analysis result. And, the evaluation model is determined according to each dimension of the evaluation model and at least one influencing factor under each dimension of the evaluation model. Among them, each dimension of the evaluation model includes at least one procurement evaluation dimension whose correlation with the project to be evaluated is higher than the first preset threshold value among each dimension of the multiple procurement evaluation dimensions in the procurement professional model. At least one influencing factor under each dimension of the evaluation model is at least one influencing factor whose influence degree on the procurement result of the project to be evaluated is greater than the second preset threshold value among the multiple influencing factors under multiple dimensions of the multiple procurement evaluation dimensions in the procurement professional model.

[0134] In some embodiments, the evaluation model determination module 903 is further configured to: determine the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model. And, determine the evaluation model according to each dimension of the evaluation model, at least one influencing factor under each dimension of the evaluation model, and the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model.

[0135] In some embodiments, when determining the weight proportion of each influencing factor in at least one influencing factor under each dimension of the evaluation model, the evaluation model determination module 903 is set to: use the Fourier transform method, the random forest regression method, and the spatiotemporal joint model method to construct a weight evaluation model for the evaluation model, and obtain three weight evaluation models. The three weight evaluation models are verified according to the verification data set to obtain the verification results. And, according to the verification results, the weight proportion of each influencing factor in at least one influencing factor under each dimension of the evaluation model is determined. Among them, the verification data set is obtained by manual verification based on the preprocessed data to be evaluated.

[0136] In some embodiments, the three weight evaluation models are verified according to the verification data set, and when the verification results are obtained, the evaluation model determination module 903 is set to: obtain the baseline performance F1 value of each influencing factor under each dimension of the three weight evaluation models according to the verification data set. And, randomly arrange the influencing factors under each dimension of the three weight evaluation models, and record the performance F1 value of each influencing factor under each dimension of the three weight evaluation models after scrambling. Among them, the verification results include the baseline performance F1 value of each influencing factor under each dimension of the three weight evaluation models and the performance F1 value of each influencing factor under each dimension of the three weight evaluation models after scrambling. .

[0137] In some embodiments, when determining the weight proportion of each influencing factor in at least one influencing factor under each dimension of the evaluation model according to the verification results, the evaluation model determination module 903 is set to: respectively compare the performance F1 value of each influencing factor under each dimension of the three shuffled weight evaluation models with the benchmark performance F1 value of each influencing factor under each dimension of the three weight evaluation models, and obtain the difference results of each influencing factor under each dimension of the three weight evaluation models. And, determine the weight proportion of each influencing factor in at least one influencing factor under each dimension of the evaluation model according to the difference results of each influencing factor under each dimension of the three weight evaluation models.

[0138] In some embodiments, the performance evaluation module 904 is configured to input at least one influencing factor under each dimension of the model to be evaluated, data corresponding to each influencing factor in at least one influencing factor under each dimension of the evaluation model in the project to be evaluated, and the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model into the procurement professional big model, so that the procurement professional big model can summarize the project to be evaluated by using text comprehension and text generation capabilities, and output performance evaluation results, improvement measures and optimization strategies.

[0139] In some embodiments, the evaluation model determination module 903 is further configured to: input at least one influencing factor under each dimension of the evaluation model, data corresponding to each influencing factor in at least one influencing factor under each dimension of the evaluation model in similar projects, and the weight proportion of each influencing factor in at least one influencing factor under each dimension of the evaluation model into the procurement professional large model, optimize the evaluation model according to the project results of similar projects and the output results of the procurement professional large model, and obtain an optimized evaluation model. Among them, similar projects are historical projects similar to the project to be evaluated. In this case, the performance evaluation module 904 is configured to: perform performance evaluation on the project to be evaluated according to the optimized evaluation project and the procurement professional large model.

[0140] In some embodiments, Fig.10 As shown, the performance evaluation system 900 also includes a module 905 for acquiring a large model of procurement expertise. The module 905 for acquiring a large model of procurement expertise is configured to: acquire original data. Structural processing is performed on the original data to obtain data after structural processing. Conversion preprocessing is performed on the data after structural processing to obtain data after conversion preprocessing. Fine-tune the parameters of the preset large model to obtain a fine-tuned large model. And, the fine-tuned large model is trained using the data after conversion preprocessing to obtain a large model of procurement expertise. The original data includes procurement expertise documents and data of historical procurement projects.

[0141] In some embodiments, when the data after structured processing is converted and preprocessed, the acquisition module 905 of the procurement professional large model is set to: use embedding technology to extract the meaning of each sentence in the data after structured processing, calculate the correlation between each sentence, obtain multiple data blocks according to the correlation between each sentence, perform knowledge slicing on each data block in the multiple data blocks to obtain sliced ​​data, use the term frequency-inverse file frequency TF-IDF algorithm to extract keywords from the sliced ​​data, and perform vectorization conversion and vector entry on the sliced ​​data to obtain the data after conversion preprocessing.

[0142] The following is a specific example to illustrate an efficiency evaluation system based on a large procurement professional model provided by an embodiment of the present solution.

[0143] like Fig.11 In the performance evaluation system flow chart of this example based on the procurement professional model, the data source is first constructed, and then the data set is constructed based on the data source to prepare for the performance evaluation. Asset management adds business descriptions to the data, and automatically generates a performance evaluation report based on the evaluation model.

[0144] The system is mainly divided into 6 modules:

[0145] 1. Data source: Before performance evaluation, you need to prepare business data of various dimensions and create a connected data source to connect to the database. It supports databases such as MySql and Oracle, as well as big data platforms such as Hadoop and Spark. It also supports BONC's XCloud and data platform.

[0146] 2. Dataset: Prepare data for project performance evaluation. According to the needs of various dimensional factors, you can establish a dataset model by selecting a table or associating an existing dataset, or by writing SQL or uploading Excel to create a dataset model.

[0147] 3. Efficiency Workbench: Provides energy efficiency analysis views from national and provincial perspectives, allowing you to intuitively view the comprehensive analysis of the efficiency of each project.

[0148] 4. Modeling Center: Manage custom models, general models, and benchmark models. Edit the evaluation dimensions, influencing factors, and weight ratio of each factor in the evaluation model.

[0149] 6. Performance dashboard: Customize performance evaluation reports based on the evaluation of each dimension of the project. You can select the evaluation report output template to quickly create reports. At the same time, you can also quickly configure various analyses through advanced configurations such as parameter configuration and theme switching.

[0150] like Fig.12 As shown, the system is mainly divided into four core layers: support engine layer, data capability layer, five-dimensional model layer and application layer.

[0151] Among them, the support engine layer is the basic capability support, mainly used to provide data, that is, the data source for performance evaluation. The support engine layer includes portal engine, knowledge engine, process engine, data engine, integration engine, intelligent engine and monitoring engine.

[0152] The data capability layer is mainly responsible for data governance, and can perform data aggregation, data processing, data expansion, statistical analysis and security monitoring.

[0153] The five-dimensional model layer is used to provide five dimensions and influencing factors under the five dimensions for the evaluation project. The five dimensions are procurement timeliness, competition effectiveness, economic efficiency, contract performance guarantee and procurement compliance. Among them, the influencing factors of procurement timeliness include: procurement preparation, procurement implementation, contract signing, and one-time success in recruitment. The influencing factors of competition effectiveness include: the number of applicants / number of companies participating in the review, the number of qualified companies, technical and commercial competition, price competition, etc. The influencing factors of economic efficiency include: capital surplus, low price for technology proximity, low price for technology superiority, and rejection of the lowest price. The influencing factors of contract performance guarantee include: supplier post-evaluation, on-time delivery, contractually agreed delivery time, and delivery time after the supplier places an order. The influencing factors of procurement compliance include: electronic bid evaluation room, subjective review weight, announcement time, and expert extraction.

[0154] The application layer is used for performance analysis and performance report output. The application layer can provide national project analysis and evaluation dashboards, procurement project energy efficiency analysis, custom project evaluation models, project energy efficiency evaluation reports and other applications.

[0155] Through the performance evaluation system based on the procurement professional big model in this example, multi-source data linking can be achieved: it supports multiple data sources and data set links, and quickly builds performance evaluation reports based on the created business data. Five-dimensional model analysis technology can be performed: the performance evaluation model analysis comprehensively evaluates and analyzes from five dimensions: competitive effectiveness, procurement timeliness, economic benefits, contract performance guarantee, and procurement compliance. The principal component analysis method, hierarchical analysis method and procurement professional big model are used to combine each influencing factor to determine the evaluation weight, and score and standardize the evaluation for each dimension. Performance evaluation report: a rich performance evaluation report template library, dynamically associated evaluation business and indicator data, indicators and statistical analysis in the report can jump to the project performance evaluation function for fine-grained drill-down analysis. And it can visualize BI: it supports business personnel in various cities, districts, counties, grids, etc. to dynamically link business data, and automatically convert data into visualization forms such as graphics, charts and dashboards to help business personnel better understand, analyze and use performance evaluation data.

[0156] An embodiment of the present invention provides a computer device, such as Fig.13 As shown, the computer device 1300 includes a memory 1301 and a processor 1302. The memory 1301 stores a computer program. When the processor 1302 runs the computer program stored in the memory 1301, the processor 1302 executes the above-mentioned performance evaluation method based on the procurement professional large model.

[0157] The specific scheme and beneficial effects of a computer device provided by an embodiment of the present invention can refer to the relevant description of a performance evaluation method based on a large procurement professional model provided by an embodiment of the present invention, which will not be repeated here.

[0158] An embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the processor executes the above-mentioned performance evaluation method based on the procurement professional large model.

[0159] The specific scheme and beneficial effects of a computer-readable storage medium provided by an embodiment of the present invention can be referred to the relevant description of a performance evaluation method based on a procurement professional large model provided by an embodiment of the present invention, which will not be repeated here.

[0160] The specific scheme and beneficial effects of a performance evaluation system based on a large model of procurement profession provided by an embodiment of the present invention can be referred to the relevant description of a performance evaluation method based on a large model of procurement profession provided by an embodiment of the present invention, and will not be repeated here.

[0161] It is to be understood that the above embodiments are merely exemplary embodiments used to illustrate the principles of the present invention, but the present invention is not limited thereto. For those of ordinary skill in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A performance evaluation method based on a large procurement professional model, characterized in that: include: Acquire data of the project to be evaluated, wherein the data of the project to be evaluated includes a purchase order, a purchase project, purchase implementation, purchase results, bidding information, contract information, and approval information of the project to be evaluated; Preprocessing the data of the project to be evaluated to obtain preprocessed data to be evaluated; Determine the evaluation model based on the data to be evaluated and the preset procurement professional model; as well as The performance of the project to be evaluated is evaluated based on the evaluation model and the procurement professional model.

2. The efficiency evaluation method based on the procurement professional large model according to claim 1 is characterized in that: The preprocessing of the data of the project to be evaluated to obtain the preprocessed data to be evaluated includes: Using the BART model, a transformer model with information denoising and bidirectional language modeling and autoregressive mechanism, to perform denoising, structuring and cleaning on the data of the project to be evaluated, and obtain the preliminarily processed data; and The preliminarily processed data is automatically labeled by a BERT-base model, a bidirectional encoder representation model from a converter, to obtain automatically labeled data, and the preprocessed data to be evaluated includes the automatically labeled data.

3. The efficiency evaluation method based on the procurement professional large model according to claim 2 is characterized in that: Preprocessing the data of the project to be evaluated to obtain preprocessed data to be evaluated, further comprising: Manual verification is performed on the automatically labeled data to obtain manually verified labeled data, and the preprocessed data to be evaluated includes the manually verified labeled data.

4. The efficiency evaluation method based on the procurement professional large model according to claim 2 is characterized in that: The automatically labeling the preliminarily processed data by using the BERT-base model includes: One or more intent tags are assigned to the preliminarily processed data according to an intent recognition algorithm.

5. The efficiency evaluation method based on the procurement professional large model according to claim 1 is characterized in that: Determining the evaluation model based on the data to be evaluated and the procurement professional model includes: Inputting the data to be evaluated into the procurement professional model; The correlation between each dimension of the multiple procurement evaluation dimensions in the procurement professional model and the data to be evaluated is identified by calculating the Pearson correlation coefficient, and each dimension of the evaluation model is determined according to the correlation identification result; each dimension of the evaluation model includes at least one procurement evaluation dimension among each dimension of the multiple procurement evaluation dimensions in the procurement professional model, the correlation with the project to be evaluated being higher than a first preset threshold value; Analyzing the influence degree of each of the multiple influencing factors under each dimension of the multiple procurement evaluation dimensions in the procurement professional model on the procurement result of the project to be evaluated by using a nonlinear regression analysis algorithm and the data to be evaluated, and determining at least one influencing factor under each dimension of the evaluation model according to the analysis result; the at least one influencing factor under each dimension of the evaluation model includes at least one influencing factor of the multiple influencing factors under each dimension of the multiple procurement evaluation dimensions in the procurement professional model, the influence degree of which on the procurement result of the project to be evaluated is greater than a second preset threshold; and The evaluation model is determined according to each dimension of the evaluation model and at least one influencing factor under each dimension of the evaluation model.

6. The efficiency evaluation method based on the procurement professional big model according to claim 5 is characterized in that: Determining the evaluation model according to each dimension of the evaluation model and at least one influencing factor under each dimension of the evaluation model includes: Determining the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model; and The evaluation model is determined according to each dimension of the evaluation model, at least one influencing factor under each dimension of the evaluation model, and a weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model.

7. The efficiency evaluation method based on the procurement professional big model according to claim 6 is characterized in that: Determining the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model includes: Fourier transform method, random forest regression method and spatiotemporal joint model method are used to construct weight evaluation models for the evaluation model, respectively, to obtain three weight evaluation models; Verifying the three weight evaluation models respectively according to the verification data set to obtain verification results; the verification data set is obtained by manually verifying the pre-processed data to be evaluated; and The weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model is determined according to the verification result.

8. The efficiency evaluation method based on the procurement professional big model according to claim 7 is characterized in that: The three weight evaluation models are verified respectively according to the verification data set to obtain verification results, including: Obtaining the benchmark performance F1 value of each influencing factor under each dimension of the three weight evaluation models respectively according to the verification data set; and The influencing factors under each dimension of the three weight evaluation models are randomly arranged respectively, and the performance F1 values ​​of the influencing factors under each dimension of the three weight evaluation models after the shuffle are recorded; the verification results include the baseline performance F1 values ​​of the influencing factors under each dimension of the three weight evaluation models and the performance F1 values ​​of the influencing factors under each dimension of the three weight evaluation models after the shuffle.

9. The efficiency evaluation method based on the procurement professional large model according to claim 8 is characterized in that: Determining the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model according to the verification result includes: Respectively comparing the performance F1 values ​​of each influencing factor under each dimension of the three weighted evaluation models after the shuffle with the benchmark performance F1 values ​​of each influencing factor under each dimension of the three weighted evaluation models to obtain difference results of each influencing factor under each dimension of the three weighted evaluation models; and The weight proportion of each influencing factor in at least one influencing factor under each dimension of the evaluation model is determined according to the difference results of each influencing factor under each dimension of the three weight evaluation models.

10. The efficiency evaluation method based on the procurement professional big model according to claim 1 is characterized in that: The performance evaluation of the project to be evaluated based on the evaluation model and the procurement professional model includes: At least one influencing factor under each dimension of the evaluation model, data corresponding to each influencing factor in at least one influencing factor under each dimension of the evaluation model in the project to be evaluated, and the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model are input into the procurement professional big model, so that the procurement professional big model can summarize the project to be evaluated by using text comprehension and text generation capabilities, and output performance evaluation results, improvement measures and optimization strategies.

11. The efficiency evaluation method based on the procurement professional big model according to claim 1 is characterized in that: Also includes: Input at least one influencing factor under each dimension of the evaluation model, data corresponding to each influencing factor in at least one influencing factor under each dimension of the evaluation model in similar projects, and the weight ratio of each influencing factor in at least one influencing factor under each dimension of the evaluation model into the procurement professional big model, optimize the evaluation model according to the project results of the similar projects and the output results of the procurement professional big model, and obtain an optimized evaluation model; wherein the similar projects are historical projects similar to the project to be evaluated; The performing of performance evaluation on the project to be evaluated based on the evaluation model and the procurement professional large model includes: performing performance evaluation on the project to be evaluated based on the optimized evaluation project and the procurement professional large model.

12. The efficiency evaluation method based on the procurement professional big model according to claim 1 is characterized in that: The method for acquiring the professional large model includes: Acquiring original data; the original data includes procurement expertise documents and data of historical procurement projects; Performing structured processing on the original data to obtain structured processed data; Performing conversion preprocessing on the data after the structured processing to obtain data after conversion preprocessing; Fine-tune the parameters of the preset large model to obtain a fine-tuned large model; and The fine-tuned large model is trained using the converted preprocessed data to obtain a procurement professional large model.

13. The efficiency evaluation method based on the procurement professional big model according to claim 11 is characterized in that: The step of performing conversion preprocessing on the data after the structured processing comprises: The embedding technology is used to extract the meaning of each sentence in the data after the structured processing, and the correlation between each sentence is calculated. According to the correlation between each sentence, multiple data blocks are obtained, and knowledge slicing is performed on each data block in the multiple data blocks to obtain sliced ​​data. The word frequency-inverse document frequency TF-IDF algorithm is used to extract keywords from the sliced ​​data, and the sliced ​​data is vectorized and vectorized to obtain the data after conversion preprocessing.

14. A performance evaluation system based on a large procurement professional model, characterized in that: include: The data acquisition module of the project to be evaluated is configured to: acquire the data of the project to be evaluated, wherein the data of the project to be evaluated includes the purchase order, purchase project, purchase implementation, purchase result, bidding information, contract information and approval information of the project to be evaluated; A preprocessing module is configured to: preprocess the data of the item to be evaluated to obtain preprocessed data to be evaluated; An evaluation model determination module is configured to: determine an evaluation model according to the data to be evaluated and a preset procurement professional model; as well as The performance evaluation module is configured to: perform performance evaluation on the project to be evaluated based on the evaluation model and the procurement professional model.

15. A computer device, characterized in that: It includes a memory and a processor, wherein the memory stores a computer program, and when the processor runs the computer program stored in the memory, the processor executes the performance evaluation method based on the procurement professional big model according to any one of claims 1 to 13.

16. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the processor executes the effectiveness evaluation method based on the procurement professional big model according to any one of claims 1 to 13.

Images