Software function point evaluation method and system based on large model

Abstract

The invention relates to the technical field of information processing, and discloses a software function point evaluation method and system based on a large model, and the method comprises the steps: carrying out the preprocessing operation of text data in a preset enterprise research and development efficiency training library, obtaining a preprocessed corpus, initializing the model parameters of a pre-training model, and obtaining the model parameters of the pre-training model; and inputting the labeled data set in the corpus to the pre-training model for iterative training to obtain a trained target model, inputting a to-be-tested data set corresponding to the function point of the to-be-tested software to the target model for calculation to obtain a function point scale estimation parameter and an enterprise production efficiency parameter, and calculating the function point scale estimation parameter and the enterprise production efficiency parameter. And according to the function point scale estimation parameter, the enterprise production efficiency parameter and a preset cost estimation algorithm, calculating to obtain an estimated cost of the to-be-tested software. It can be seen that the enterprise software cost estimation process and the quantitative standard can be unified by implementing the method, the applicability of the software cost estimation scheme is improved, and therefore the efficiency and accuracy of software cost estimation can be improved.

Description

Technical Field

[0001] The present invention relates to the field of information processing technology, and in particular to a software function point evaluation method and system based on a large model. Background Art

[0002] In the current traditional software cost estimation scheme, the calculation method of traditional software cost estimation is generally based on the scale of the function points of the object being evaluated, the average production efficiency of the industry and the average labor cost of the region to calculate the final cost. However, this calculation method often has the following shortcomings: First of all, the industry average production efficiency and regional average labor cost are statistical data, which are quite different from the enterprises themselves and often do not have much reference value, especially for small and micro enterprises and large enterprises.

[0003] Secondly, the degree of reuse of traditional software cost estimation methods only considers the software being evaluated, and does not consider whether the company has undertaken similar projects, whether it has accumulated similar controls, components or core resources of product lines. Therefore, using traditional estimation methods, if the upgrade project is undertaken by different manufacturers, the estimated cost will be lower than the actual production, and for companies with rich resource accumulation, the estimated cost will be higher, and the company cannot accurately assess its actual production cost.

[0004] In addition, traditional software cost estimation methods often do not consider the matching degree between the development language, technical route and the company's own technical reserves. For situations where the specific development language and specific technical route that meet the project requirements do not match the company's own technical reserves, the cost estimated by the traditional estimation method will show obvious deviations, and the manual evaluation process is relatively slow. When dealing with large-scale or complex software projects, the efficiency and accuracy of the evaluation are low.

[0005] Therefore, a software function point evaluation method and system based on a large model is currently provided, which can unify the process and quantitative standards of enterprise software cost estimation, improve the applicability of software cost estimation solutions, and realize the intelligence and automation of software cost estimation, which is conducive to improving the efficiency and accuracy of software cost evaluation. Summary of the invention

[0006] The present invention provides a method and system for software function point evaluation based on a large model, which can standardize the process and method of enterprise software cost estimation, improve the applicability of software cost estimation solutions, and thus help improve the efficiency and accuracy of software cost evaluation.

[0007] In order to solve the above technical problems, the first aspect of the present invention discloses a software function point evaluation method based on a large model, the method comprising: Performing a preprocessing operation on the text data in the preset enterprise R&D efficiency training library to obtain a preprocessed corpus, wherein the corpus at least includes an annotated data set corresponding to the software function points with annotated evaluation parameters; Initializing the model parameters of the pre-trained model, and inputting the annotated data set in the corpus into the pre-trained model for iterative training to obtain a trained target model; Inputting the test data set corresponding to the function points of the software to be tested into the target model for calculation to obtain function point scale estimation parameters and enterprise production efficiency parameters; The estimated cost of the software to be tested is calculated based on the function point scale estimation parameters, the enterprise production efficiency parameters, the enterprise cost parameters and a preset cost estimation algorithm.

[0008] As an optional implementation, in the first aspect of the present invention, the enterprise R&D efficiency training library includes: at least one of enterprise human resources data, enterprise revenue and expenditure data, historical project demand data, historical project technical route data, and historical project development architecture data; the annotated data set includes a plurality of words and / or phrases and / or short sentences for describing the software function points; and the preprocessing operation includes: at least one of data cleaning, data filtering, data conversion, and data normalization processing; Before initializing the model parameters of the pre-trained model, the method further includes: Constructing a loss function of the pre-trained model based on the characters and / or words and / or sentences in the corpus; The calculation formula of the loss function is:

[0009]

[0010] Wherein, the coppus means that the expected library includes a plurality of words and / or phrases and / or short sentences for describing the software function points, is represented as the loss function, p is represented as the probability function, is represented by the nth word, phrase or sentence in the corpus, Represented as the number of tokens in the corpus, Represented as GPT model parameters; And, after initializing the model parameters of the pre-trained model, the method further includes: Freeze some model parameters of the pre-trained model according to preset freezing parameters.

[0011] As an optional implementation, in the first aspect of the present invention, after the text data in the preset enterprise R&D efficiency training library is preprocessed to obtain the preprocessed corpus, the method further includes: According to the semantic type matching the text data corresponding to each software function point in the corpus, the text data corresponding to each software function point is annotated with a semantic type; and / or, According to the grammatical type matched with the text data corresponding to each software function point in the corpus, the text data corresponding to each software function point is annotated with a grammatical type; And, according to the preset evaluation parameters, determining the evaluation parameters that match the text data corresponding to each software function point, and obtaining the annotated data set corresponding to the software function point annotated with the evaluation parameters; The evaluation parameters include: at least one of a development benefit level, a development cost level, a development scale level, and a complexity level that matches the text data corresponding to each software function point.

[0012] As an optional implementation, in the first aspect of the present invention, the pre-trained model is represented as a sequence-to-sequence neural network model based on a self-attention mechanism, and the model parameters of the pre-trained model at least include a self-attention layer, a multi-head attention layer, and a feedforward neural network layer; The self-attention layer includes: input sequence, query, key, value, attention weight, output layer. The input sequence of the self-attention layer is expressed as , the parameter of the input sequence is the coppus, and the i represents the i-th element of the input sequence; The linear transformation calculation formula of the query, the key, and the value is:

[0013] The calculation formula of the attention weight is:

[0014] The calculation formula of the output layer is:

[0015] Wherein, Q represents query, K represents key, V represents value, and Represented as a query weight matrix, Expressed as a key weight matrix, is represented as a value weight matrix, the attention is represented as the attention weight, the d is represented as the dimension, and the output is represented as the output parameter; The multi-head attention layer includes: the input sequence, the head, the attention weight of the head, and the output layer; The calculation formula for the attention weight of each head is:

[0016] The calculation formula of the output layer is:

[0017] Among them, the represents the i-th head of the multi-head attention, h represents the number of the heads, selfAttention represents the self-attention mechanism function, WO represents the linear transformation corresponding to the output layer of the self-attention layer, and multihead_attention represents the output parameter of the multi-head attention layer; The feedforward neural network layer at least includes: an input layer, a hidden layer, and an output layer.

[0018] As an optional implementation, in the first aspect of the present invention, the step of inputting the annotated data set in the corpus into the pre-trained model for iterative training to obtain a trained target model comprises: Inputting the annotated data set in the corpus into the pre-trained model for iterative training to obtain a trained model; Determining whether the trained model tends to converge according to the loss function of the pre-trained model; When it is determined that the trained model has not converged, the model parameters are updated according to the calculated gradient of the loss function and by an optimization algorithm, and the iterative training operation is repeated on the trained model; When it is determined that the trained model tends to converge, the iterative training operation is stopped to obtain the final target model.

[0019] As an optional implementation, in the first aspect of the present invention, the optimization algorithm includes: one of gradient descent, stochastic gradient descent, mini-batch gradient descent, Adam, RMSProp, momentum, and AdaGrad; The preset cost estimation algorithm is expressed as: .

[0020] As an optional implementation, in the first aspect of the present invention, the enterprise production efficiency parameter includes: at least one of a reuse coefficient, a complexity coefficient, an industry coefficient, and a cost coefficient; The step of inputting the test data set corresponding to the function points of the software to be tested into the target model for calculation to obtain the function point scale estimation parameters and the enterprise production efficiency parameters includes: Inputting the test data set corresponding to the function points of the software to be tested into the target model for calculation to obtain function point scale estimation parameters and historical project characteristic parameters; Enterprise production efficiency parameters are generated based on the function point scale estimation parameters and historical project characteristic parameters.

[0021] The second aspect of the present invention discloses a software function point evaluation system based on a large model, the system comprising: A preprocessing module is used to perform a preprocessing operation on the text data in the preset enterprise R&D efficiency training library to obtain a preprocessed corpus, wherein the corpus at least includes an annotated data set corresponding to the software function points with annotated evaluation parameters; Initialization module, used to initialize the model parameters of the pre-trained model; A training module, used for inputting the annotated data set in the corpus obtained by the preprocessing module into the pre-trained model initialized by the initialization module for iterative training to obtain a trained target model; A calculation module is used to input the test data set corresponding to the function points of the software to be tested into the target model trained by the training module for calculation, so as to obtain function point scale estimation parameters and enterprise production efficiency parameters; and to obtain the estimated cost of the software to be tested based on the function point scale estimation parameters, the enterprise production efficiency parameters, the enterprise cost parameters and a preset cost estimation algorithm.

[0022] As an optional implementation, in the second aspect of the present invention, the enterprise R&D efficiency training library includes: at least one of enterprise human resources data, enterprise revenue and expenditure data, historical project demand data, historical project technical route data, and historical project development architecture data; the annotated data set includes a plurality of words and / or phrases and / or short sentences for describing the software function points; the preprocessing operation includes: at least one of data cleaning, data filtering, data conversion, and data normalization processing; The system further comprises: The construction module is used to construct the loss function of the pre-training model according to the characters and / or words and / or short sentences in the corpus before the initialization module initializes the model parameters of the pre-training model; The calculation formula of the loss function is:

[0023]

[0024] Wherein, the coppus means that the expected library includes a plurality of words and / or phrases and / or short sentences for describing the software function points, is represented as the loss function, p is represented as the probability function, is represented by the nth word, phrase or sentence in the corpus, Represented as the number of tokens in the corpus, Represented as GPT model parameters; And, a freezing module is used to freeze part of the model parameters of the pre-trained model according to preset freezing parameters after the initialization module initializes the model parameters of the pre-trained model.

[0025] As an optional implementation, in the second aspect of the present invention, the system further includes: A labeling module, used for performing a preprocessing operation on the text data in a preset enterprise R&D efficiency training library in the preprocessing module, and after obtaining the preprocessed corpus, labeling the text data corresponding to each software function point in the corpus with a semantic type according to the semantic type matching the text data corresponding to each software function point in the corpus; and / or labeling the text data corresponding to each software function point in the corpus with a grammatical type according to the grammatical type matching the text data corresponding to each software function point in the corpus; and, according to preset evaluation parameters, determining the evaluation parameters matching the text data corresponding to each software function point, and obtaining a labeling data set corresponding to the software function point labeled with the evaluation parameters; The evaluation parameters include: at least one of a development benefit level, a development cost level, a development scale level, and a complexity level that matches the text data corresponding to each software function point.

[0026] As an optional implementation, in the second aspect of the present invention, the pre-trained model is represented as a sequence-to-sequence neural network model based on a self-attention mechanism, and the model parameters of the pre-trained model at least include a self-attention layer, a multi-head attention layer, and a feedforward neural network layer; The self-attention layer includes: input sequence, query, key, value, attention weight, output layer. The input sequence of the self-attention layer is expressed as , the parameter of the input sequence is the coppus, and the i represents the i-th element of the input sequence; The linear transformation calculation formula of the query, the key, and the value is:

[0027] The calculation formula of the attention weight is:

[0028] The calculation formula of the output layer is:

[0029] Wherein, Q represents query, K represents key, V represents value, and Represented as a query weight matrix, Expressed as a key weight matrix, is represented as a value weight matrix, the attention is represented as the attention weight, the d is represented as the dimension, and the output is represented as the output parameter; The multi-head attention layer includes: the input sequence, the head, the attention weight of the head, and the output layer; The calculation formula for the attention weight of each head is:

[0030] The calculation formula of the output layer is:

[0031] Among them, the represents the i-th head of the multi-head attention, h represents the number of the heads, selfAttention represents the self-attention mechanism function, WO represents the linear transformation corresponding to the output layer of the self-attention layer, and multihead_attention represents the output parameter of the multi-head attention layer; The feedforward neural network layer at least includes: an input layer, a hidden layer, and an output layer.

[0032] As an optional implementation, in the second aspect of the present invention, the training module inputs the annotated data set in the corpus into the pre-trained model for iterative training, and the specific method of obtaining the trained target model is: Inputting the annotated data set in the corpus into the pre-trained model for iterative training to obtain a trained model; Determining whether the trained model tends to converge according to the loss function of the pre-trained model; When it is determined that the trained model has not converged, the model parameters are updated according to the calculated gradient of the loss function and by an optimization algorithm, and the iterative training operation is repeated on the trained model; When it is determined that the trained model tends to converge, the iterative training operation is stopped to obtain the final target model.

[0033] As an optional implementation, in the second aspect of the present invention, the optimization algorithm includes: one of gradient descent, stochastic gradient descent, mini-batch gradient descent, Adam, RMSProp, momentum, and AdaGrad; The preset cost estimation algorithm is expressed as: .

[0034] As an optional implementation, in the second aspect of the present invention, the enterprise production efficiency parameter includes: at least one of a reuse coefficient, a complexity coefficient, an industry coefficient, and a cost coefficient; The calculation module inputs the test data set corresponding to the function points of the software to be tested into the target model for calculation, and obtains the function point scale estimation parameters and the enterprise production efficiency parameters in the following specific manner: Inputting the test data set corresponding to the function points of the software to be tested into the target model for calculation to obtain function point scale estimation parameters and historical project characteristic parameters; Enterprise production efficiency parameters are generated based on the function point scale estimation parameters and historical project characteristic parameters.

[0035] The third aspect of the present invention discloses another software function point evaluation system based on a large model, the system comprising: A memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the software function point evaluation method based on the large model disclosed in the first aspect of the present invention.

[0036] The fourth aspect of the present invention discloses a computer storage medium, wherein the computer storage medium stores computer instructions, and when the computer instructions are called, they are used to execute the software function point evaluation method based on a large model disclosed in the first aspect of the present invention.

[0037] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: The present invention provides a software function point evaluation method and system based on a large model, the method comprising: preprocessing text data in a preset enterprise R&D efficiency training library to obtain a preprocessed corpus, the corpus at least including a labeled data set corresponding to the software function points with annotated evaluation parameters, which is conducive to improving the efficiency and accuracy of subsequent pre-model training; initializing the model parameters of the pre-training model, and inputting the labeled data set in the corpus into the pre-training model for iterative training to obtain a trained target model, which is conducive to improving the accuracy of software cost evaluation; inputting the test data set corresponding to the function points of the software to be tested into the target model for calculation, obtaining function point scale estimation parameters and enterprise production efficiency parameters, and calculating the estimated cost of the software to be tested according to the function point scale estimation parameters, enterprise production efficiency parameters and a preset cost estimation algorithm. It can be seen that the implementation of the present invention can unify the process and quantitative standards of enterprise software cost estimation, improve the applicability of the software cost estimation scheme, realize the intelligence and automation of software cost estimation, reduce the time for manual review of documents and the time required for the cost process, thereby helping to improve the efficiency and accuracy of software cost evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0039] Figure 1 It is a flowchart of a software function point evaluation method based on a large model disclosed in an embodiment of the present invention; Figure 2 It is a flow chart of another software function point evaluation method based on a large model disclosed in an embodiment of the present invention; Figure 3 It is a structural diagram of a software function point evaluation system based on a large model disclosed in an embodiment of the present invention; Figure 4 It is a structural schematic diagram of another software function point evaluation system based on a large model disclosed in an embodiment of the present invention; Figure 5 It is a structural diagram of another software function point evaluation system based on a large model disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0040] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only 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.

[0041] The terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish different objects rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or terminal including a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units inherent to these processes, methods, products or terminals.

[0042] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0043] The present invention discloses a software function point evaluation method and system based on a large model, which can perform preprocessing operations on text data in a preset enterprise R&D efficiency training library to obtain a preprocessed corpus, which at least includes annotated data sets corresponding to software function points with annotated evaluation parameters, which is conducive to improving the efficiency and accuracy of subsequent pre-model training; initialize the model parameters of the pre-training model, and input the annotated data set in the corpus into the pre-training model for iterative training to obtain a trained target model, input the data set to be tested corresponding to the function points of the software to be tested into the target model for calculation, obtain function point scale estimation parameters and enterprise production efficiency parameters, and finally calculate the estimated cost of the software to be tested according to the function point scale estimation parameters, enterprise production efficiency parameters and a preset cost estimation algorithm, which can unify the process and quantitative standards of enterprise software cost estimation, improve the applicability of software cost estimation schemes, realize the intelligence and automation of software cost estimation, reduce the time for manual review of documents and the time required for the cost process, thereby helping to improve the efficiency and accuracy of software cost evaluation. The following are detailed descriptions.

[0044] Embodiment 1 See also Figure 1 , Figure 1 : is a flow chart of a software function point evaluation method based on a large model disclosed in an embodiment of the present invention. Figure 1 The described software function point evaluation method based on a large model can be applied to a software function point evaluation system based on a large model, and can also be applied to a software cost estimation system based on a large model, and the embodiments of the present invention do not limit this. Optionally, the system can be applied to a local terminal device (e.g., a PC terminal, a local server, etc.) or a mobile terminal device (e.g., a smart phone, a tablet computer, a PDA, a mobile Internet device, etc.), and the embodiments of the present invention do not limit this. Figure 1 As shown, the software function point evaluation method based on the large model may include the following operations: 101. Perform preprocessing operations on the text data in the preset enterprise R&D efficiency training library to obtain a preprocessed corpus.

[0045] In an embodiment of the present invention, optionally, the enterprise R&D efficiency training library may include: at least one of enterprise human resources data, enterprise revenue and expenditure data, historical project demand data, historical project technical route data, historical project development architecture data, etc. The corpus may at least include annotated data sets corresponding to software function points with annotated evaluation parameters, wherein the annotated data sets may include multiple words and / or phrases and / or short sentences for describing the software function points. The preprocessing operation may include: at least one of data cleaning, data filtering, data conversion, data normalization, etc., wherein the data conversion operation may include: semantic conversion and / or translation, grammatical conversion and / or simplification of text data, for example: for the same software function point (user login function), the text description of the software manual A is that the system supports user login, while the text description of the software manual B is that the system supports user registration, and the registered account is used for login, and the login error is prompted, etc. This situation is actually an equivalent function, so it needs to be converted into a standard text description through a data conversion operation, which is not limited in the embodiment of the present invention. This provides a unified reference basic standard, which is conducive to providing text reference accuracy, thereby improving the efficiency and accuracy of software cost evaluation.

[0046] 102. Initialize the model parameters of the pre-trained model, and input the labeled data set in the corpus into the pre-trained model for iterative training to obtain a trained target model.

[0047] In an embodiment of the present invention, optionally, the pre-trained model is represented as a sequence-to-sequence neural network model (Transformer model) based on a self-attention mechanism, and the model parameters of the pre-trained model may at least include a self-attention layer, a multi-head attention layer, a feedforward neural network layer, etc.

[0048] Among them, the attention layer can include: input sequence, query, key, value, attention weight, output layer, and the input sequence of the self-attention layer is expressed as , the parameter of the input sequence is coppus, and i represents the i-th element of the input sequence.

[0049] The linear transformation calculation formula for query, key, and value is:

[0050] The calculation formula of attention weight is:

[0051] The calculation formula of the output layer is:

[0052] Among them, Q represents query, K represents key, and V represents value. Represented as the query weight matrix, Expressed as a key weight matrix, Represented as a value weight matrix, It is represented as the transposed matrix of the key, attention is represented as the attention weight, softmax is represented as the normalized function, d is represented as the dimension, and output is represented as the output parameter.

[0053] In an embodiment of the present invention, optionally, a multi-head attention layer may include: an input sequence, a head, an attention weight of the head, and an output layer.

[0054] The calculation formula for the attention weight of each head is:

[0055] The calculation formula of the output layer is:

[0056] in, It represents the i-th head of multi-head attention, h represents the number of heads, selfAttention represents the self-attention mechanism function, WO represents the linear transformation corresponding to the output layer of the self-attention layer, multihead_attention represents the output parameter of the multi-head attention layer, and concat represents the function used to connect two or more arrays.

[0057] In an embodiment of the present invention, optionally, the feedforward neural network layer may include at least: an input layer, a hidden layer, and an output layer, wherein the parameters input to the input layer of the feedforward neural network layer are the output parameters of the multi-head attention.

[0058] The calculation formula of the hidden layer is expressed as:

[0059] The calculation formula of the output layer is expressed as:

[0060] in, Represents the weight value of the hidden layer, Represents the bias value of the hidden layer, Represents the weight value of the output layer, Represents the bias value of the output layer.

[0061] In this way, by initializing the model parameters of the pre-trained Transformer model and inputting the annotated data set in the corpus into the pre-trained model for iterative training, the trained target model is obtained, which fully utilizes the advantages of high computational efficiency and strong scalability of the Transformer model, thereby helping to improve the efficiency and accuracy of software cost assessment.

[0062] In an embodiment of the present invention, inputting the annotated data set in the corpus into the pre-training model for iterative training to obtain a trained target model may include: The annotated data set in the corpus is input into the pre-trained model for iterative training to obtain the trained model.

[0063] The loss function of the pre-trained model is used to determine whether the trained model tends to converge.

[0064] When it is determined that the trained model has not converged, the model parameters are updated according to the calculated gradient of the loss function and the optimization algorithm, and the iterative training operation of the trained model is repeated.

[0065] When it is determined that the trained model tends to converge, the iterative training operation is stopped to obtain the final target model.

[0066] Among them, the optimization algorithm may include: gradient descent, stochastic gradient descent, mini-batch gradient descent, Adam, RMSProp, momentum, AdaGrad, etc., which is not limited in the embodiment of the present invention.

[0067] In this way, it is judged whether the trained model tends to converge according to the loss function of the pre-trained model. When it is judged that the trained model does not tend to converge, the model parameters are updated according to the calculated gradient of the loss function and the optimization algorithm, and the trained model is repeatedly iterated. When it is judged that the trained model tends to converge, the iterative training operation is stopped to obtain the final target model, which can prevent the training model from overfitting and can use the optimization algorithm to update the model parameters, which is conducive to accelerating the training of the target model, thereby helping to improve the efficiency and accuracy of subsequent software cost evaluation.

[0068] 103. Input the test data set corresponding to the function points of the software to be tested into the target model for calculation to obtain the function point scale estimation parameters and the enterprise production efficiency parameters, and calculate the estimated cost of the software to be tested based on the function point scale estimation parameters, the enterprise production efficiency parameters, the enterprise cost parameters and the preset cost estimation algorithm.

[0069] In the embodiment of the present invention, optionally, the enterprise production efficiency parameters may include: at least one of: reuse coefficient, complexity coefficient, industry coefficient, cost coefficient, etc., which is not limited in the embodiment of the present invention.

[0070] The preset cost estimation algorithm is expressed as:

[0071] In an embodiment of the present invention, inputting the test data set corresponding to the function points of the software to be tested into the target model for calculation to obtain the function point scale estimation parameters and the enterprise production efficiency parameters may include: The test data set corresponding to the function points of the software to be tested is input into the target model for calculation to obtain the function point scale estimation parameters and historical project characteristic parameters.

[0072] The historical project characteristic parameters can be understood as high-dimensional projection of the historical projects, or vectorization, or target characteristic values ​​calculated by the MD5 algorithm. The historical project characteristic parameters can be used to quickly identify whether a similar project has been done before a new project.

[0073] In this way, the estimated cost of the software to be tested can be calculated based on the function point scale estimation parameters, enterprise production efficiency parameters, enterprise cost parameters calculated according to the target model and the preset cost estimation algorithm. This can unify the process and quantitative standards of enterprise software cost estimation, improve the applicability of software cost estimation solutions, and thus help improve the efficiency and accuracy of software cost evaluation.

[0074] Generate enterprise production efficiency parameters based on function point size estimation parameters and historical project characteristic parameters.

[0075] In an optional embodiment, before executing the operation of step 102, the method further includes the following operations: The loss function of the pre-trained model is constructed based on the words and / or terms and / or sentences in the corpus.

[0076] In this optional embodiment, optionally, the loss function is calculated as follows:

[0077]

[0078] Among them, coppus means that the expectation library includes multiple words and / or phrases and / or short sentences used to describe software functional points. Expressed as the loss function, Represents the nth word, phrase or sentence in the corpus. Represented as the number of tokens in the corpus, Represented as GPT model parameters, the GPT model (Generative Pre-trained Transformer) is a generative pre-trained language model developed by OpenAI based on the Transformer architecture. It can understand and generate natural language text through unsupervised pre-training and supervised fine-tuning. p is represented as a probability function, which is used to calculate the probability of predicting the nth word, phrase or sentence given the context of the first K words.

[0079] It can be seen that this optional embodiment can construct the loss function of the pre-trained model based on the text and / or words and / or short sentences in the corpus, reduce the error of the training model, thereby preventing overfitting of the training model, and is conducive to improving the efficiency and accuracy of software cost evaluation.

[0080] In an optional embodiment, after performing the operation of initializing the model parameters of the pre-trained model in step 102, the method further includes the following operations: Freeze some model parameters of the pre-trained model according to the preset freezing parameters. For example, you can freeze some invalid or unused weight values ​​in the hidden layer of the pre-trained model according to the specific type of data in the test data set.

[0081] It can be seen that this optional embodiment can freeze some model parameters of the pre-trained model according to preset freezing parameters, thereby improving the computational efficiency and accuracy of the model and preventing overfitting of the training model, which is beneficial to improving the efficiency and accuracy of software cost evaluation.

[0082] Embodiment 2 See also Figure 2 , Figure 2 : is a flow chart of a software function point evaluation method based on a large model disclosed in an embodiment of the present invention. Figure 2 The described software function point evaluation method based on a large model can be applied to a software function point evaluation system based on a large model, and can also be applied to a software cost estimation system based on a large model, and the embodiments of the present invention do not limit this. Optionally, the system can be applied to a local terminal device (e.g., a PC terminal, a local server, etc.) or a mobile terminal device (e.g., a smart phone, a tablet computer, a PDA, a mobile Internet device, etc.), and the embodiments of the present invention do not limit this. Figure 2As shown, the software function point evaluation method based on the large model may include the following operations: 201. Perform preprocessing operations on text data in a preset enterprise R&D efficiency training library to obtain a preprocessed corpus.

[0083] 202. According to the semantic type matching the text data corresponding to each software function point in the corpus, the text data corresponding to each software function point is annotated with a semantic type. 203. According to the grammatical type matching the text data corresponding to each software function point in the corpus, the text data corresponding to each software function point is annotated with a grammatical type.

[0084] 204. Determine the evaluation parameters that match the text data corresponding to each software function point according to the preset evaluation parameters, and obtain the annotated data set corresponding to the software function point annotated with the evaluation parameters.

[0085] In an embodiment of the present invention, optionally, the evaluation parameters may include: at least one of a development benefit level, a development cost level, a development scale level, a complexity level, etc. that matches the text data corresponding to each software function point, and the embodiment of the present invention is not limited thereto.

[0086] In the embodiment of the present invention, optionally, the evaluation parameters matching each code snippet corresponding to each software function point can be determined according to the preset code evaluation parameters, and the annotated data set corresponding to the software function point annotated with the evaluation parameters can be obtained. The code evaluation parameters may include at least one of the readability, maintainability, time complexity, reusability, security, consistency, etc. of the code snippet, which is not limited in the embodiment of the present invention.

[0087] 205. Initialize the model parameters of the pre-trained model, and input the labeled data set in the corpus into the pre-trained model for iterative training to obtain a trained target model.

[0088] 206. Input the test data set corresponding to the function points of the test software into the target model for calculation to obtain the function point scale estimation parameters and the enterprise production efficiency parameters, and calculate the estimated cost of the test software based on the function point scale estimation parameters, the enterprise production efficiency parameters, the enterprise cost parameters and the preset cost estimation algorithm.

[0089] In the embodiment of the present invention, for other descriptions of step 201 and step 205-step 206, please refer to the detailed description of step 101-step 104 in the first embodiment, and the embodiment of the present invention will not be repeated.

[0090] It can be seen that the implementation of the embodiments of the present invention can annotate the text data corresponding to each software function point with the semantic type and / or grammatical type according to the semantic type and / or grammatical type matching the text data corresponding to each software function point in the corpus, and determine the evaluation parameters matching the text data corresponding to each software function point according to the preset evaluation parameters, so as to obtain the annotated data set corresponding to the software function points with the annotated evaluation parameters, which is beneficial to improving the data accuracy and diversity of the model training set, improving the training efficiency of the model, thereby improving the generalization ability of the model, and is beneficial to improving the efficiency and accuracy of software cost evaluation.

[0091] It can be seen that the implementation Figure 2 The described software function point evaluation method based on a large model can perform preprocessing operations on text data in a preset enterprise R&D efficiency training library to obtain a preprocessed corpus, which at least includes a labeled data set corresponding to the software function points with labeled evaluation parameters, which is beneficial to improving the efficiency and accuracy of subsequent pre-model training; initialize the model parameters of the pre-trained model, and input the labeled data set in the corpus into the pre-trained model for iterative training to obtain a trained target model, input the test data set corresponding to the function points of the software to be tested into the target model for calculation, obtain function point scale estimation parameters and enterprise production efficiency parameters, and finally calculate the function point scale estimation parameters, enterprise production efficiency parameters and the preset cost estimation algorithm to obtain the target model. The estimated cost of software can unify the process and quantitative standards of enterprise software cost estimation, and improve the applicability of software cost estimation solutions; and according to the semantic type and / or grammatical type matching the text data corresponding to each software function point in the corpus, the text data corresponding to each software function point is annotated with the semantic type and / or grammatical type, and according to the preset evaluation parameters, the evaluation parameters matching the text data corresponding to each software function point are determined to obtain the annotated data set corresponding to the software function point with the annotated evaluation parameters, which is conducive to improving the data accuracy and diversity of the model training set, improving the training efficiency of the model, thereby helping to improve the generalization ability of the model, and helping to improve the efficiency and accuracy of software cost evaluation.

[0092] Embodiment 3 See also Figure 3 , Figure 3 : is a schematic diagram of a software function point evaluation system based on a large model disclosed in an embodiment of the present invention. Figure 3The described software function point evaluation system based on a large model can execute the above-mentioned software function point evaluation method based on a large model, and the method can also be applied to a software cost estimation system based on a large model, which is not limited in the embodiments of the present invention. Optionally, the system can be applied to a local terminal device (e.g., a PC terminal, a local server, etc.) or a mobile terminal device (e.g., a smart phone, a tablet computer, a PDA, a mobile Internet device, etc.), which is not limited in the embodiments of the present invention. Figure 3 As shown, the software function point evaluation system based on the large model may include a preprocessing module 301, an initialization module 302, a training module 303, and a calculation module 304, wherein: The preprocessing module 301 is used to perform preprocessing operations on the text data in the preset enterprise R&D efficiency training library to obtain a preprocessed corpus, which at least includes annotated data sets corresponding to software function points with annotated evaluation parameters.

[0093] The initialization module 302 is used to initialize the model parameters of the pre-trained model.

[0094] The training module 303 is used to input the annotated data set in the corpus processed by the preprocessing module 301 into the pre-trained model initialized by the initialization module 302 for iterative training to obtain a trained target model.

[0095] The calculation module 304 is used to input the test data set corresponding to the function point of the test software into the target model trained by the training module 303 for calculation, and obtain the function point scale estimation parameters and enterprise production efficiency parameters. And, according to the function point scale estimation parameters, enterprise production efficiency parameters, enterprise cost parameters and the preset cost estimation algorithm, the estimated cost of the test software is calculated.

[0096] It can be seen that the implementation Figure 3The described large model-based software function point evaluation system can perform preprocessing operations on text data in a preset enterprise R&D efficiency training library to obtain a preprocessed corpus, which at least includes annotated data sets corresponding to software function points with annotated evaluation parameters, which is beneficial to improving the efficiency and accuracy of subsequent pre-model training; initialize the model parameters of the pre-trained model, and input the annotated data sets in the corpus into the pre-trained model for iterative training to obtain a trained target model, input the data sets to be tested corresponding to the function points of the software to be tested into the target model for calculation, and obtain function point scale estimation parameters and enterprise production efficiency parameters; finally, the estimated cost of the software to be tested is calculated based on the function point scale estimation parameters, enterprise production efficiency parameters and a preset cost estimation algorithm, which can unify the process and quantitative standards of enterprise software cost estimation, improve the applicability of software cost estimation solutions, realize the intelligence and automation of software cost measurement, reduce the time for manual document review and the time required for the cost estimation process, and thus help improve the efficiency and accuracy of software cost evaluation.

[0097] In an optional embodiment, the enterprise R&D efficiency training library includes: at least one of enterprise human resources data, enterprise income and expenditure data, historical project demand data, historical project technology route data, and historical project development architecture data; the annotated data set includes multiple texts and / or words and / or short sentences used to describe software functional points; and the preprocessing operations include: at least one of data cleaning, data filtering, data conversion, and data normalization.

[0098] As well as Figure 4 As shown, the system also includes: The construction module 305 is used to construct a loss function of the pre-training model according to the characters and / or words and / or short sentences in the corpus before the initialization module 302 initializes the model parameters of the pre-training model.

[0099] Among them, the calculation formula of the loss function is:

[0100]

[0101] Among them, coppus means that the expectation library includes multiple words and / or phrases and / or short sentences used to describe software functional points. It is represented as the loss function, p is represented as the probability function, Represents the nth word, phrase or sentence in the corpus. Represented as the number of tokens in the corpus, Represented as GPT model parameters.

[0102] And, the freezing module 306 is used to freeze part of the model parameters of the pre-trained model according to preset freezing parameters after the initialization module 302 initializes the model parameters of the pre-trained model.

[0103] It can be seen that the implementation Figure 4 The described large model-based software function point evaluation system can construct the loss function of the pre-trained model according to the text and / or words and / or short sentences in the corpus, thereby reducing the error of the training model and preventing the overfitting of the training model; and can freeze some model parameters of the pre-trained model according to preset freezing parameters, thereby improving the computational efficiency and accuracy of the model and preventing the overfitting of the training model, which is conducive to improving the efficiency and accuracy of software cost evaluation.

[0104] In another optional embodiment, Figure 4 As shown, the system also includes: The annotation module 307 is used to perform preprocessing operations on the text data in the preset enterprise R&D efficiency training library in the preprocessing module 301, and after obtaining the preprocessed corpus, perform semantic type annotation on the text data corresponding to each software function point in the corpus according to the semantic type matching the text data corresponding to each software function point in the corpus; and / or, perform grammatical type annotation on the text data corresponding to each software function point in the corpus according to the grammatical type matching the text data corresponding to each software function point in the corpus; and, according to the preset evaluation parameters, determine the evaluation parameters that match the text data corresponding to each software function point, and obtain the annotation data set corresponding to the software function point with the annotated evaluation parameters.

[0105] The evaluation parameters include: at least one of a development benefit level, a development cost level, a development scale level, and a complexity level that matches the text data corresponding to each software function point.

[0106] It can be seen that the implementation Figure 4 The described large model-based software function point evaluation system can annotate the text data corresponding to each software function point with semantic type and / or grammatical type according to the semantic type and / or grammatical type matching the text data corresponding to each software function point in the corpus, and determine the evaluation parameters matching the text data corresponding to each software function point according to preset evaluation parameters, so as to obtain the annotated data set corresponding to the software function points with annotated evaluation parameters, which is conducive to improving the data accuracy and diversity of the model training set, improving the training efficiency of the model, thereby improving the generalization ability of the model, and is conducive to improving the efficiency and accuracy of software cost evaluation.

[0107] In another optional embodiment, the pre-trained model is represented as a sequence-to-sequence neural network model based on a self-attention mechanism, and the model parameters of the pre-trained model include at least a self-attention layer, a multi-head attention layer, and a feedforward neural network layer.

[0108] Among them, the self-attention layer includes: input sequence, query, key, value, attention weight, output layer. The input sequence of the self-attention layer is expressed as , the parameter of the input sequence is coppus, and i represents the i-th element of the input sequence.

[0109] The linear transformation calculation formula for query, key, and value is:

[0110] The calculation formula of attention weight is:

[0111] The calculation formula of the output layer is:

[0112] Among them, Q represents query, K represents key, and V represents value. Represented as the query weight matrix, Expressed as a key weight matrix, It is represented as a value weight matrix, attention is represented as the attention weight, d is represented as the dimension, and output is represented as the output parameter.

[0113] The multi-head attention layer includes: input sequence, head, head attention weight, and output layer.

[0114] The calculation formula for the attention weight of each head is:

[0115] The calculation formula of the output layer is:

[0116] in, It represents the i-th head of multi-head attention, h represents the number of heads, selfAttention represents the self-attention mechanism function, WO represents the linear transformation corresponding to the output layer of the self-attention layer, and multihead_attention represents the output parameter of the multi-head attention layer.

[0117] The feedforward neural network layer includes at least: input layer, hidden layer, and output layer.

[0118] It can be seen that the implementation Figure 4The described large model-based software function point evaluation system can fully utilize the advantages of high computational efficiency and strong scalability of the Transformer model, thereby helping to improve the efficiency and accuracy of software cost evaluation.

[0119] In yet another optional embodiment, Figure 4 As shown, the training module 303 inputs the annotated data set in the corpus into the pre-training model for iterative training, and the specific method of obtaining the trained target model is: The annotated data set in the corpus is input into the pre-trained model for iterative training to obtain the trained model.

[0120] The loss function of the pre-trained model is used to determine whether the trained model tends to converge.

[0121] When it is determined that the trained model has not converged, the model parameters are updated according to the calculated gradient of the loss function and the optimization algorithm, and the iterative training operation of the trained model is repeated.

[0122] When it is determined that the trained model tends to converge, the iterative training operation is stopped to obtain the final target model.

[0123] It can be seen that the implementation Figure 4 The described large-model-based software function point evaluation system can determine whether the trained model tends to converge according to the loss function of the pre-trained model. When it is determined that the trained model does not tend to converge, the model parameters are updated according to the calculated gradient of the loss function and through the optimization algorithm, and the trained model is repeatedly iteratively trained; when it is determined that the trained model tends to converge, the iterative training operation is stopped to obtain the final target model, which can prevent the training model from overfitting, and can use the optimization algorithm to update the model parameters, which is conducive to accelerating the training of the target model, thereby facilitating improving the efficiency and accuracy of subsequent software cost evaluation.

[0124] In another optional embodiment, the optimization algorithm includes: one of gradient descent, stochastic gradient descent, mini-batch gradient descent, Adam, RMSProp, momentum, and AdaGrad.

[0125] The preset cost estimation algorithm is expressed as: .

[0126] It can be seen that the implementation Figure 4The described large model-based software function point evaluation system can calculate the estimated cost of the software to be tested according to the function point scale estimation parameters calculated by the target model, the enterprise production efficiency parameters, the enterprise cost parameters and the preset cost estimation algorithm. It can unify the process and quantitative standards of enterprise software cost estimation, improve the applicability of software cost estimation solutions, and thus help improve the efficiency and accuracy of software cost evaluation.

[0127] In yet another optional embodiment, the enterprise production efficiency parameters include: at least one of a reuse coefficient, a complexity coefficient, an industry coefficient, and a cost coefficient.

[0128] As well as Figure 4 As shown, the calculation module 304 inputs the test data set corresponding to the function points of the test software into the target model for calculation, and obtains the function point scale estimation parameters and the enterprise production efficiency parameters in the specific manner as follows: The test data set corresponding to the function points of the software to be tested is input into the target model for calculation to obtain the function point scale estimation parameters and historical project characteristic parameters.

[0129] Generate enterprise production efficiency parameters based on function point size estimation parameters and historical project characteristic parameters.

[0130] It can be seen that the implementation Figure 4 The described large model-based software function point evaluation system can calculate the estimated cost of the software to be tested according to the function point scale estimation parameters calculated by the target model, the enterprise production efficiency parameters, the enterprise cost parameters and the preset cost estimation algorithm. It can unify the process and quantitative standards of enterprise software cost estimation, improve the applicability of software cost estimation solutions, and thus help improve the efficiency and accuracy of software cost evaluation.

[0131] Embodiment 4 See also Figure 5 , Figure 5 FIG. 1 is a schematic diagram of another software function point evaluation system based on a large model disclosed in an embodiment of the present invention. Figure 5 As shown, the software function point evaluation system based on the large model may include: A memory 401 storing executable program codes; a processor 402 coupled to the memory 401; The processor 402 calls the executable program code stored in the memory 401 to execute the steps in the software function point evaluation method based on the large model described in the first embodiment of the present invention or the second embodiment of the present invention.

[0132] Embodiment 5 An embodiment of the present invention discloses a computer storage medium, which stores computer instructions. When the computer instructions are called, they are used to execute the steps in the software function point evaluation method based on a large model described in Embodiment 1 or Embodiment 2 of the present invention.

[0133] Embodiment 6 An embodiment of the present invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to enable a computer to execute the steps in the large model-based software function point evaluation method described in Example 1 or Example 2.

[0134] The system embodiment described above is only illustrative, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, that is, they may be located in one place, or they may be distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without paying creative labor.

[0135] Through the specific description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on such an understanding, the above technical solution can be essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable rewritable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0136] Finally, it should be noted that the software function point evaluation method and system based on a large model disclosed in the embodiment of the present invention only discloses the preferred embodiment of the present invention, which is only used to illustrate the technical solution 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 the technical solutions described in the aforementioned embodiments can still be modified, or some of the technical features can be replaced by equivalents. 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 various embodiments of the present invention.

Claims

1. A software function point evaluation method based on a large model, characterized in that: The method comprises: Performing a preprocessing operation on the text data in the preset enterprise R&D efficiency training library to obtain a preprocessed corpus, wherein the corpus at least includes an annotated data set corresponding to the software function points with annotated evaluation parameters; Initializing the model parameters of the pre-trained model, and inputting the annotated data set in the corpus into the pre-trained model for iterative training to obtain a trained target model; Inputting the test data set corresponding to the function points of the software to be tested into the target model for calculation to obtain function point scale estimation parameters and enterprise production efficiency parameters; The estimated cost of the software to be tested is calculated based on the function point scale estimation parameters, the enterprise production efficiency parameters, the enterprise cost parameters and a preset cost estimation algorithm.

2. The software function point evaluation method based on a large model according to claim 1 is characterized in that: The enterprise R&D efficiency training library includes: at least one of enterprise human resources data, enterprise revenue and expenditure data, historical project demand data, historical project technical route data, and historical project development architecture data; the annotated data set includes a plurality of words and / or phrases and / or short sentences for describing the software function points; the preprocessing operation includes: at least one of data cleaning, data filtering, data conversion, and data normalization processing; Before initializing the model parameters of the pre-trained model, the method further includes: Constructing a loss function of the pre-trained model based on the characters and / or words and / or sentences in the corpus; The calculation formula of the loss function is: Wherein, the coppus means that the expected library includes a plurality of words and / or phrases and / or short sentences for describing the software function points, is represented as the loss function, p is represented as the probability function, is represented by the nth word, phrase or sentence in the corpus, Represented as the number of tokens in the corpus, Represented as GPT model parameters; And, after initializing the model parameters of the pre-trained model, the method further includes: Freeze some model parameters of the pre-trained model according to preset freezing parameters.

3. The software function point evaluation method based on a large model according to claim 2 is characterized in that: After the text data in the preset enterprise R&D efficiency training library is preprocessed to obtain the preprocessed corpus, the method further includes: According to the semantic type matching the text data corresponding to each software function point in the corpus, the text data corresponding to each software function point is annotated with a semantic type; and / or, According to the grammatical type matched with the text data corresponding to each software function point in the corpus, the text data corresponding to each software function point is annotated with a grammatical type; And, according to the preset evaluation parameters, determining the evaluation parameters that match the text data corresponding to each software function point, and obtaining the annotated data set corresponding to the software function point annotated with the evaluation parameters; The evaluation parameters include: at least one of a development benefit level, a development cost level, a development scale level, and a complexity level that matches the text data corresponding to each software function point.

4. The software function point evaluation method based on a large model according to claim 2 is characterized in that: The pre-trained model is represented as a sequence-to-sequence neural network model based on a self-attention mechanism, and the model parameters of the pre-trained model include at least a self-attention layer, a multi-head attention layer, and a feedforward neural network layer; The self-attention layer includes: input sequence, query, key, value, attention weight, output layer, and the input sequence of the self-attention layer is expressed as , the parameter of the input sequence is the coppus, and the i represents the i-th element of the input sequence; The linear transformation calculation formula of the query, the key, and the value is: The calculation formula of the attention weight is: The calculation formula of the output layer is: Wherein, Q represents query, K represents key, V represents value, and Represented as a query weight matrix, Expressed as a key weight matrix, is represented as a value weight matrix, the attention is represented as the attention weight, the d is represented as the dimension, and the output is represented as the output parameter; The multi-head attention layer includes: the input sequence, the head, the attention weight of the head, and the output layer; The calculation formula for the attention weight of each head is: The calculation formula of the output layer is: Among them, the represents the i-th head of the multi-head attention, h represents the number of the heads, selfAttention represents the self-attention mechanism function, WO represents the linear transformation corresponding to the output layer of the self-attention layer, and multihead_attention represents the output parameter of the multi-head attention layer; The feedforward neural network layer includes at least: an input layer, a hidden layer, and an output layer.

5. The software function point evaluation method based on a large model according to claim 2 is characterized in that: The step of inputting the annotated data set in the corpus into the pre-trained model for iterative training to obtain a trained target model includes: Inputting the annotated data set in the corpus into the pre-trained model for iterative training to obtain a trained model; Determining whether the trained model tends to converge according to the loss function of the pre-trained model; When it is determined that the trained model has not converged, the model parameters are updated according to the calculated gradient of the loss function and by an optimization algorithm, and the iterative training operation is repeated on the trained model; When it is determined that the trained model tends to converge, the iterative training operation is stopped to obtain the final target model.

6. The software function point evaluation method based on a large model according to claim 5 is characterized in that: The optimization algorithm includes: one of gradient descent, stochastic gradient descent, mini-batch gradient descent, Adam, RMSProp, momentum, and AdaGrad; The calculation formula of the cost estimation algorithm is expressed as: 。 7. The software function point evaluation method based on a large model according to claim 5 is characterized in that: The enterprise production efficiency parameters include: at least one of a reuse coefficient, a complexity coefficient, an industry coefficient, and a cost coefficient; The step of inputting the test data set corresponding to the function points of the software to be tested into the target model for calculation to obtain the function point scale estimation parameters and the enterprise production efficiency parameters includes: Inputting the test data set corresponding to the function points of the software to be tested into the target model for calculation to obtain function point scale estimation parameters and historical project characteristic parameters; Enterprise production efficiency parameters are generated based on the function point scale estimation parameters and historical project characteristic parameters.

8. A software function point evaluation system based on a large model, characterized in that: The system comprises: A preprocessing module is used to perform a preprocessing operation on the text data in the preset enterprise R&D efficiency training library to obtain a preprocessed corpus, wherein the corpus at least includes an annotated data set corresponding to the software function points with annotated evaluation parameters; Initialization module, used to initialize the model parameters of the pre-trained model; A training module, used for inputting the annotated data set in the corpus obtained by the preprocessing module into the pre-trained model initialized by the initialization module for iterative training to obtain a trained target model; A calculation module is used to input the test data set corresponding to the function points of the software to be tested into the target model trained by the training module for calculation, so as to obtain function point scale estimation parameters and enterprise production efficiency parameters; and to obtain the estimated cost of the software to be tested based on the function point scale estimation parameters, the enterprise production efficiency parameters, the enterprise cost parameters and a preset cost estimation algorithm.

9. A software function point evaluation system based on a large model, characterized in that: The system comprises: A memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the software function point evaluation method based on a large model as described in any one of claims 1-7.

10. A computer storage medium, characterized in that: The computer storage medium stores computer instructions, and when the computer instructions are called, they are used to execute the software function point evaluation method based on a large model as described in any one of claims 1-7.

Images