Technology development system based on artificial intelligence

Abstract

The invention discloses a technology development system based on artificial intelligence, which belongs to the technical field of technology development and comprises an information acquisition module, an information understanding module, a function keyword weight similarity judgment module, a function keyword coding module, a technology development code homogenization judgment module and a database. According to the technical development completed according to the natural language description of the user about the to-be-realized function of technical development, the user does not need to have perfect technical development capability, homogenization of technical development results can be avoided, and the market competitiveness of completed technical development finished products can be improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of technology development, and in particular relates to a technology development system based on artificial intelligence. Background Art

[0002] AI-based technology development systems refer to the use of AI theories, methods, and techniques to develop new computer systems or improve existing systems so that they have higher intelligence levels and stronger adaptability. These systems can perform complex tasks such as learning, reasoning, decision-making, and natural language processing, thereby improving work efficiency, reducing costs, and creating new value.

[0003] Chinese Patent No. 202311364807.9 discloses a technology development system and method based on artificial intelligence technology, including: obtaining a natural language description of a computer function module to be implemented; performing word segmentation processing on the natural language description of the computer function module to be implemented to obtain a sequence of function module description words; passing the sequence of function module description words through a word embedding layer to obtain a sequence of function module description word embedding vectors; passing the sequence of function module description word embedding vectors through a first semantic encoder based on a Transformer module to obtain a plurality of function module description word feature vectors; arranging the plurality of function module description word feature vectors into a two-dimensional feature matrix and then optimizing them to obtain an optimized two-dimensional feature matrix; passing the optimized two-dimensional feature matrix through a second semantic encoder based on a convolutional neural network model to obtain a function module semantic understanding feature vector; and passing the function module semantic understanding feature vector through a code generator based on a natural language processing model to generate a computer code.

[0004] The above technology has the following problems: the computer code of the system is generated by the coder of the natural language processing model based on the optimized semantic understanding feature vector. The coder generates computer code based on the optimized semantic understanding feature vector generally based on the preset semantic understanding feature vector-computer code rules. The preset semantic understanding feature vector-computer code rules generally correspond to the same semantic understanding feature vector and the same computer code. This leads to serious homogeneity when users use the system to generate computer code to assist in technology development, reducing the market competitiveness of user technology development.

[0005] In view of this, a technology development system based on artificial intelligence is designed to solve the above problems. Summary of the invention

[0006] To solve the problems raised in the above background technology. The present invention provides a technology development system based on artificial intelligence, which has the characteristics of completing technology development according to the user's natural language description of the functions to be realized in technology development, not requiring the user to have perfect technology development capabilities, but also avoiding the homogenization of technology development results, and improving the market competitiveness of the completed technology development products.

[0007] To achieve the above object, the present invention provides the following technical solution: a technology development system based on artificial intelligence, comprising:

[0008] Information acquisition module, which obtains the user's natural language description of the functions to be implemented in technology development;

[0009] The information understanding module performs word segmentation on the acquired user's natural language description and extracts keywords for the functions to be implemented in technology development;

[0010] A function keyword weight similarity determination module determines the weight similarity between the extracted technical development function keywords to be implemented and the existing user technical development function keywords to be implemented;

[0011] A function keyword coding module determines the coding rules of function keywords based on the function keyword weight similarity. If the function keyword weight similarity between two users is lower than a preset threshold, the function keyword coding model can be used to encode according to any coding rule corresponding to the function keyword to generate a computer code. Otherwise, the coding rule corresponding to the function keyword and the coding rule used by the existing user function keyword are avoided. The function keyword coding model can be used to encode according to the remaining coding rules corresponding to the function keyword to generate a computer code, and the computer code of the function keyword is integrated to generate a technical development code that can realize the user function.

[0012] The technical development code homogeneity determination module performs a running test based on the generated technical development code that can realize the user function. During the test, the homogeneity between the technical development code and the existing user technical development code is determined. If the homogeneity is lower than the preset threshold, the technical development is completed. Otherwise, the function keyword encoding module is returned to avoid the encoding rules with a homogeneity higher than the preset threshold based on the function keyword encoding model and re-encodes. The above steps are repeated until the homogeneity is lower than the preset threshold.

[0013] The database stores the function keyword weights and codes of the technical development completed by the user for use in the above comparative analysis.

[0014] Furthermore, the function keyword weight of the function keyword weight similarity determination module is related to the position in the natural language description. The closer the position of the function keyword in the natural language description, the greater the weight. This rule prompts the user before the information acquisition module acquires the natural language description.

[0015] Furthermore, the weight similarity formula of the function keyword weight similarity determination module is:

[0016]

[0017] In the formula: a represents user a, b represents user b, k represents the function keyword, It is represented as the weight of the kth function keyword of user a among all function keywords. It is represented as the weight of the kth function keyword of user b among all function keywords.

[0018] Furthermore, the function keyword encoding model of the function keyword encoding module is a pre-trained model, and the specific steps are as follows:

[0019] Constructing a function keyword-based coding model, the function keyword-based coding model includes an input layer, a plurality of parallel coding layers, a plurality of parallel coding scheme layers, a fusion coding layer, a fusion coding scheme layer and an output layer, wherein the input layer is connected to the plurality of parallel coding layers one by one, the plurality of parallel coding layers are connected to the corresponding plurality of parallel coding scheme layers one by one, the plurality of parallel coding layers are connected to the fusion coding layer one by one, the fusion coding layer is connected to the fusion coding scheme layer, and the fusion coding layer is connected to the output layer;

[0020] Collecting historical function keywords, computer codes of the function keywords, and fusion technology development codes of the computer codes as a data set;

[0021] Divide the dataset into training and testing sets;

[0022] By training and testing the function keyword coding model based on data in the training set and the test set, the coding scheme is optimized until convergence. Then, the function keywords can be encoded and integrated based on the function keyword coding model to generate technical development codes that can realize user functions.

[0023] Furthermore, the homogeneity degree formula in the technical development code homogeneity determination module is:

[0024]

[0025] Where: a represents user a, b represents user b, X a It represents all the functional expressions of the technical development code of user a in the running test, Xb It represents all functional expressions of the technical development code of user b in the running test, and N represents all users.

[0026] Compared with the prior art, the present invention has the following beneficial effects:

[0027] The present invention first obtains the user's natural language description of the function to be implemented in the technical development, then performs word segmentation on the natural language description, extracts the function keywords, and then determines the encoding rules based on the function keywords based on the full-center similarity comparison between the function keywords and the existing user function keywords, and then encodes the function keywords based on the determined encoding rules to generate computer codes, and then fuses the computer codes to generate the overall technical development code, and then determines the homogeneity based on the test run of the overall technical development code, and then determines the completion of the technical development or re-encoding based on the determined homogeneity, until the technical development is completed. Compared with the existing technology, the technical development completed according to the user's natural language description of the function to be implemented in the technical development does not require the user to have perfect technical development capabilities, but can avoid the homogeneity of technical development results, and can improve the market competitiveness of the completed technical development products. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 A framework diagram of the technology development system based on artificial intelligence of the present invention;

[0029] In the figure: 1. Information acquisition module; 2. Information understanding module; 3. Function keyword weight similarity judgment module; 4. Function keyword encoding module; 5. Technology development code homogeneity judgment module; 6. Database. DETAILED DESCRIPTION

[0030] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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.

[0031] See also Figure 1 , the present invention provides the following technical solution: a technology development system based on artificial intelligence, comprising:

[0032] Information acquisition module 1, obtaining the user's natural language description of the function to be implemented in the technical development;

[0033] Before obtaining the user's natural language description of the function to be implemented in the technical development, and before the user triggers the input instruction of the natural language description of the function to be implemented in the technical development, the user is first prompted that the position before and after the natural language description affects the weight of the function keyword. The earlier the position, the higher the weight of the function keyword, and vice versa;

[0034] When obtaining the user's natural language description of the function to be implemented by the technical development, the user enters the natural language description in order according to the required importance, that is, weight, of the function to be implemented by the technical development;

[0035] Information understanding module 2, segmenting the acquired user natural language description and extracting keywords for functions to be implemented in technology development;

[0036] Remove insignificant characters from the natural language descriptions obtained from users, such as extra spaces and line breaks, as well as high-frequency but meaningless words such as "is" and "in";

[0037] Use the HanLP word segmentation tool to segment the user's natural language description;

[0038] Filter based on part-of-speech tagging, that is, extract the keywords of the functions to be realized in technology development;

[0039] Function keyword weight similarity determination module 3, determines the weight similarity between the extracted technical development function keywords to be implemented and the existing user technical development function keywords to be implemented;

[0040] Determining the weighted similarity between the keywords of the technical development functions to be implemented of the user and the existing users is performed in the order of the weights of the keywords of the technical development functions to be implemented of the users, that is, first determining the weighted similarity between the keywords of the technical development functions to be implemented of the user with the highest weight and the keywords of the technical development functions to be implemented of the existing users with the highest weight, retaining the keywords of the technical development functions to be implemented of the existing users whose weighted similarity is higher than the preset threshold, and then determining the weighted similarity between the keywords of the technical development functions to be implemented of the user with the second highest weight and the keywords of the technical development functions to be implemented of the existing users with the second highest weight, and so on, until the weighted similarities of the keywords of the technical development functions to be implemented of the users are all compared and stopped;

[0041] The function keyword coding module 4 determines the coding rules of the function keywords based on the function keyword weight similarity. If the function keyword weight similarity between two users is lower than a preset threshold, the function keyword coding model can be used to encode according to any coding rule corresponding to the function keyword to generate a computer code. Otherwise, the coding rule corresponding to the function keyword and the coding rule used by the existing user function keyword are avoided. The function keyword coding model can be used to encode according to the remaining coding rules corresponding to the function keyword to generate a computer code, and the computer code of the function keyword is integrated to generate a technical development code that can realize the user function.

[0042] The technical development code homogeneity determination module 5 performs a running test based on the generated technical development code that can realize the user function. During the test, the homogeneity between the technical development code and the existing user technical development code is determined. If the homogeneity is lower than the preset threshold, the technical development is completed. Otherwise, the function keyword encoding module 4 is returned to avoid the encoding rules with a homogeneity higher than the preset threshold based on the function keyword encoding model and re-encodes. The above steps are repeated until the homogeneity is lower than the preset threshold.

[0043] During the running test, the homogeneity determination is performed based on the functional expression formed by the computer code of one functional keyword and the functional expression of the computer code of another functional keyword. If the homogeneity degree is higher than a preset threshold, the computer code of the functional keyword is returned to the functional keyword encoding module 4 for single re-encoding instead of overall re-encoding;

[0044] Database 6 stores the function keyword weights and codes of the technical development completed by the user for use in the above-mentioned comparative analysis.

[0045] Specifically, the weight similarity formula of the function keyword weight similarity determination module 3 is:

[0046]

[0047] In the formula: a represents user a, b represents user b, k represents the function keyword, It is represented as the weight of the kth function keyword of user a among all function keywords. It is represented as the weight of the kth function keyword of user b among all function keywords.

[0048] Specifically, the function keyword encoding model of the function keyword encoding module 4 is a pre-trained model, and the specific steps are as follows:

[0049] Constructing a function keyword-based coding model, the function keyword-based coding model includes an input layer, a plurality of parallel coding layers, a plurality of parallel coding scheme layers, a fusion coding layer, a fusion coding scheme layer and an output layer, wherein the input layer is connected to the plurality of parallel coding layers one by one, the plurality of parallel coding layers are connected to the corresponding plurality of parallel coding scheme layers one by one, the plurality of parallel coding layers are connected to the fusion coding layer one by one, the fusion coding layer is connected to the fusion coding scheme layer, and the fusion coding layer is connected to the output layer;

[0050] The number of the plurality of coding layers and the plurality of coding scheme layers can be increased or decreased and correspond one to one, that is, if there is no suitable coding scheme in the plurality of coding scheme layers to encode the function keywords, the coding scheme layers are expanded based on data training, and during the expansion of the coding scheme layers, the coding scheme layers are expanded synchronously and correspond one to one;

[0051] Collecting historical function keywords, computer codes of the function keywords, and fusion technology development codes of the computer codes as a data set;

[0052] Divide the dataset into training and testing sets;

[0053] By training and testing the function keyword coding model based on data in the training set and the test set, the coding scheme is optimized until convergence. Then, the function keywords can be encoded and integrated based on the function keyword coding model to generate technical development codes that can realize user functions.

[0054] Specifically, the homogeneity degree formula in the technology development code homogeneity determination module 5 is:

[0055]

[0056] Where: a represents user a, b represents user b, X a It represents all the functional expressions of the technical development code of user a in the running test, X b It represents all functional expressions of the technical development code of user b in the running test, and N represents all users.

[0057] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A technology development system based on artificial intelligence, characterized in that: include: Information acquisition module (1), obtaining a user's natural language description of the function to be implemented in the technology development; The information understanding module (2) performs word segmentation on the acquired user natural language description and extracts keywords for functions to be implemented in technology development; Function keyword weight similarity determination module (3), determines the weight similarity between the extracted technical development function keywords to be implemented and the existing user technical development function keywords to be implemented; A function keyword encoding module (4) determines the encoding rules of the function keywords based on the function keyword weight similarity. If the function keyword weight similarity between two users is lower than a preset threshold, the function keyword encoding model can be used to encode according to any encoding rule corresponding to the function keyword to generate a computer code. Otherwise, the encoding rule corresponding to the function keyword and the encoding rule used by the existing user function keyword are avoided. The function keyword encoding model can be used to encode according to the remaining encoding rules corresponding to the function keyword to generate a computer code, and the computer code of the function keyword is integrated to generate a technical development code that can realize the user function. The technical development code homogeneity determination module (5) performs a running test based on the generated technical development code that can realize the user function. During the test, the homogeneity between the technical development code and the existing user technical development code is determined. If the homogeneity is lower than a preset threshold, the technical development is completed. Otherwise, the function keyword encoding module (4) is returned to avoid the encoding rules with a homogeneity higher than the preset threshold based on the function keyword encoding model and re-encodes. The above steps are repeated until the homogeneity is lower than the preset threshold. Database (6) stores the function keyword weights and codes of the technical development completed by the user for use in the above-mentioned comparative analysis.

2. The artificial intelligence-based technology development system according to claim 1, characterized in that: The function keyword weight of the function keyword weight similarity determination module (3) is related to the position in the natural language description. The closer the position of the function keyword in the natural language description is, the greater the weight is. This rule is prompted to the user before the information acquisition module (1) acquires the natural language description.

3. The artificial intelligence-based technology development system according to claim 1, characterized in that: The weight similarity formula of the function keyword weight similarity determination module (3) is: In the formula: a represents user a, b represents user b, k represents the function keyword, It is represented as the weight of the kth function keyword of user a among all function keywords. It is represented as the weight of the kth function keyword of user b among all function keywords.

4. The artificial intelligence-based technology development system according to claim 1, characterized in that: The function keyword encoding model of the function keyword encoding module (4) is a pre-trained model, and the specific steps are as follows: Constructing a function keyword-based coding model, the function keyword-based coding model includes an input layer, a plurality of parallel coding layers, a plurality of parallel coding scheme layers, a fusion coding layer, a fusion coding scheme layer and an output layer, wherein the input layer is connected to the plurality of parallel coding layers one by one, the plurality of parallel coding layers are connected to the corresponding plurality of parallel coding scheme layers one by one, the plurality of parallel coding layers are connected to the fusion coding layer one by one, the fusion coding layer is connected to the fusion coding scheme layer, and the fusion coding layer is connected to the output layer; Collecting historical function keywords, computer codes of the function keywords, and fusion technology development codes of the computer codes as a data set; Divide the dataset into training and testing sets; By training and testing the function keyword coding model based on data in the training set and the test set, the coding scheme is optimized until convergence. Then, the function keywords can be encoded and integrated based on the function keyword coding model to generate technical development codes that can realize user functions.

5. The artificial intelligence-based technology development system according to claim 1, characterized in that: The homogeneity degree formula in the technical development code homogeneity determination module (5) is: Where: a represents user a, b represents user b, X a It represents all the functional expressions of the technical development code of user a in the running test, X b It represents all functional expressions of the technical development code of user b in the running test, and N represents all users.

Images