A data generation method and device, electronic device and storage medium

Abstract

The present disclosure discloses a data generation method and device, an electronic device and a storage medium, constructs a target prompt word; inputs the target prompt word into a preset language model, generates a feature engineering code for extracting the target feature according to the preset language model; analyzes the data set to be analyzed according to the feature engineering code to obtain the target feature. By inputting the preconfigured prompt word containing the data set description and the target feature name into the preset language model, the feature engineering code is automatically generated through the preset language model, and then the data set to be analyzed is processed to obtain the target feature, so as to achieve the update of the feature library and provide data support for subsequent anti-fraud tasks. By automatically generating feature engineering code, the tedious process of traditional manual coding is avoided, the speed and efficiency of feature engineering are improved, and the possibility of errors caused by manual intervention is reduced, ensuring the accuracy and reliability of feature extraction, and providing a solid data foundation for subsequent anti-fraud tasks.

Description

Technical Field

[0001] The present disclosure relates to the field of data processing technology, and in particular to a data generation method and device, an electronic device, and a storage medium. Background Art

[0002] With the rapid development of technologies such as the Internet, big data, and artificial intelligence, online fraud methods are becoming increasingly innovative, posing huge risks to banks and individual users.

[0003] Traditional anti-fraud methods mostly rely on predefined feature libraries, which are often built based on historical data and are difficult to adapt to the rapid evolution of fraud methods. The perpetrators constantly change their methods and use technical means to evade detection, resulting in a continuous decline in the recognition rate of fraud by traditional anti-fraud methods. Summary of the invention

[0004] The present disclosure provides a data generation method, device, electronic device and storage medium, which is mainly intended to solve the problem that the predefined feature library is difficult to adapt to the rapid evolution of fraud methods, resulting in a continuous decline in the recognition rate of fraud by traditional anti-fraud methods.

[0005] According to a first aspect of the present disclosure, a method for generating data is provided, comprising:

[0006] Constructing a target prompt word; wherein the target prompt word includes a description of the data set to be analyzed and a target feature name;

[0007] Inputting the target prompt word into a preset language model, and generating a feature engineering code for extracting the target feature according to the preset language model;

[0008] The data set to be analyzed is analyzed according to the feature engineering code to obtain the target feature.

[0009] Optionally, the target prompt word includes at least one of the context information and semantic information of the data set to be analyzed, the target feature name and the annotation of the target feature and the expected feature engineering code.

[0010] Optionally, the construction target prompt word also includes:

[0011] Inputting the target prompt word into a training language model to generate a first training feature engineering code; wherein the first training feature engineering code has the same format as the expected feature engineering code;

[0012] Analyze the training data set according to the first training feature engineering code to obtain a first training target feature;

[0013] Determining a second training target feature having a positive performance improvement in the first training target feature;

[0014] The target prompt word is reconfigured, and training is performed according to the preset target prompt word after the reconfiguration, until the training times reach a preset times threshold, and then the training is stopped; wherein the preset configuration prompt word after the reconfiguration includes the feature name of the second training target feature.

[0015] Optionally, the step of inputting the target prompt word into a preset language model and generating a feature engineering code for extracting the target feature according to the preset language model further comprises:

[0016] The trained target prompt word is input into the preset language model, and a feature engineering code for extracting the target feature is generated according to the preset language model.

[0017] Optionally, inputting the target prompt word into a preset language model, and generating a feature engineering code for extracting the target feature according to the preset language model further comprises:

[0018] The preset language model generates an annotation for the target feature; wherein the annotation includes at least one of a feature extraction method and an applicable scenario of the target feature.

[0019] According to a second aspect of the present disclosure, there is provided a data generating device, comprising:

[0020] A construction unit, used to construct a target prompt word; wherein the target prompt word includes a description of the data set to be analyzed and a target feature name;

[0021] A generating unit, configured to input the target prompt word into a preset language model, and generate a feature engineering code for extracting the target feature according to the preset language model;

[0022] An analysis unit is used to analyze the data set to be analyzed according to the feature engineering code to obtain the target feature.

[0023] Optionally, the target prompt word includes at least one of the context information and semantic information of the data set to be analyzed, the target feature name and the annotation of the target feature and the expected feature engineering code.

[0024] Optionally, the construction unit further includes:

[0025] An input module, used for inputting the target prompt word into a training language model to generate a first training feature engineering code; wherein the first training feature engineering code has the same format as the expected feature engineering code;

[0026] an analysis module, configured to analyze the training data set according to the first training feature engineering code to obtain a first training target feature;

[0027] A determination module, configured to determine a second training target feature having a positive performance improvement in the first training target feature;

[0028] A construction module is used to reconfigure the target prompt word and train according to the preset target prompt word after the reconfiguration until the training times reach a preset times threshold, and then stop the training; wherein the preset configuration prompt word after the reconfiguration contains the feature name of the second training target feature.

[0029] Optionally, the generating unit is further used for:

[0030] The trained target prompt word is input into the preset language model, and a feature engineering code for extracting the target feature is generated according to the preset language model.

[0031] Optionally, the generating unit further includes:

[0032] The preset language model generates an annotation for the target feature; wherein the annotation includes at least one of a feature extraction method and an applicable scenario of the target feature.

[0033] According to a third aspect of the present disclosure, there is provided an electronic device, including:

[0034] at least one processor; and

[0035] a memory communicatively connected to the at least one processor; wherein,

[0036] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method described in the first aspect.

[0037] According to a fourth aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to execute the method described in the first aspect.

[0038] According to a fifth aspect of the present disclosure, a computer program product is provided, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements the method as described in the first aspect above.

[0039] The data generation method, device, electronic device and storage medium provided by the present disclosure, the main technical scheme includes: constructing a target prompt word; wherein the target prompt word includes the description of the data set to be analyzed and the target feature name; inputting the target prompt word into a preset language model, and generating a feature engineering code for extracting the target feature according to the preset language model; analyzing the data set to be analyzed according to the feature engineering code to obtain the target feature. Compared with the related art, the embodiment of the present application automatically generates feature engineering code through the preset language model by inputting the preconfigured prompt word including the data set description and the target feature name into the preset language model, and then processes the data set to be analyzed to obtain the target feature, realizes the update of the feature library, and provides data support for subsequent anti-fraud tasks. By automatically generating feature engineering code, the tedious process of traditional manual coding is avoided, the speed and efficiency of feature engineering are improved, and the possibility of errors caused by manual intervention is reduced, ensuring the accuracy and reliability of feature extraction, and providing a solid data foundation for subsequent anti-fraud tasks.

[0040] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0042] Figure 1 A flowchart of a method for generating data provided by an embodiment of the present disclosure;

[0043] Figure 2 A flowchart of a method for generating data provided by an embodiment of the present disclosure;

[0044] Figure 3 A schematic diagram of a flow chart of a method for generating a prompt word provided in an embodiment of the present disclosure;

[0045] Figure 4 A schematic diagram of the structure of a data generation device provided in an embodiment of the present disclosure;

[0046] Figure 5 A schematic diagram of the structure of a data generation device provided in an embodiment of the present disclosure;

[0047] Figure 6 A schematic block diagram of an exemplary electronic device provided for an embodiment of the present disclosure. DETAILED DESCRIPTION

[0048] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0049] The following describes a data generation method, apparatus, electronic device, and storage medium according to an embodiment of the present disclosure with reference to the accompanying drawings.

[0050] Figure 1 A flowchart of a method for generating data provided by an embodiment of the present disclosure.

[0051] like Figure 1 As shown, the method comprises the following steps:

[0052] Step 101, constructing a target prompt word; wherein the target prompt word includes a description of the data set to be analyzed and a target feature name.

[0053] A prompt is a text or instruction that provides input to a model to guide it to generate a specific output. It is a text paragraph provided by the user when interacting with the model, which is used to describe the information, answer, text, etc. that the user wants to obtain from the model. The purpose of the prompt is to guide the model to produce the desired response so that the generated output can be better controlled.

[0054] For language models, a prompt can be a short question, a full paragraph, or a set of instructions, depending on the needs and scenario. When generating text, the model will try to understand the prompt and generate a corresponding response based on its understanding. The prompt directly affects the content, style, and quality of the text generated by the model.

[0055] In some embodiments, when constructing the target prompt word, it can be constructed based on historical experience, such as knowing how to construct the target prompt word with the best effect based on historical experience; or a standard template for the target prompt word can be determined based on experiments, and the target prompt word can be generated based on the standard template; specifically, the embodiments of the present application do not limit the method for constructing the target prompt word.

[0056] Step 102: input the target prompt word into a preset language model, and generate a feature engineering code for extracting the target feature according to the preset language model.

[0057] In some embodiments, the target prompt word is input into a preset language model, which can be GPT, Gemini or other large language models, etc. The embodiment of the present application does not limit the large prediction model. The preset language model is a preset language model that has been trained with a large amount of data and can understand the semantics of the language and generate corresponding codes. The embodiment of the present application takes the preset language model as a large language model as an example for explanation, but it should be noted that this narrative method is only an exemplary description and is not a specific limitation on the preset language model.

[0058] The big prediction model analyzes relevant features based on the input target prompt words. For example, if the target feature in the target prompt word is changed to "user behavior", the big prediction model will select features such as the user's historical click count and visit duration. It also generates feature engineering code for the feature, which is used to select target features, standardize numerical features, and perform one-hot encoding on categorical features in the subsequent feature processing process so that they can be effectively input into the model for training, etc.

[0059] Step 103: Analyze the data set to be analyzed according to the feature engineering code to obtain the target feature.

[0060] In some embodiments, before executing this step, the data set to be analyzed is preprocessed. The data set to be analyzed is the user's personal data collected by the bank's internal system from multiple sources, including transaction data, account information, device information, location data, user behavior logs, etc. Each personal account will collect hundreds of dimensions of personal data, and contain time series data. In some embodiments, for example, to facilitate data storage and data transmission, a big data stream processing framework (such as ApacheKafka, Spark Streaming) is used to achieve real-time data storage and transmission to ensure the timeliness of the data.

[0061] Data preprocessing includes data cleaning and data conversion, such as removing or filling missing values, handling outliers, ensuring data consistency and accuracy; converting non-numeric data into numerical data, such as using One-Hot Encoding to process categorical variables; finally, through standardization or normalization, the numerical data has a unified standard scale.

[0062] According to the feature engineering code, feature extraction is performed on the data set to be analyzed. Feature extraction includes creating meaningful features, deleting unnecessary features, etc.

[0063] The data generation method provided by the present disclosure mainly includes the following technical solutions: constructing a target prompt word; wherein the target prompt word includes the description of the data set to be analyzed and the target feature name; inputting the target prompt word into a preset language model, and generating a feature engineering code for extracting the target feature according to the preset language model; analyzing the data set to be analyzed according to the feature engineering code to obtain the target feature. Compared with the related art, the embodiment of the present application automatically generates feature engineering code through the preset language model by inputting the preconfigured prompt word including the data set description and the target feature name into a preset language model, and then processes the data set to be analyzed to obtain the target feature, thereby updating the feature library and providing data support for subsequent anti-fraud tasks. By automatically generating feature engineering code, the tedious process of traditional manual coding is avoided, the speed and efficiency of feature engineering are improved, and the possibility of errors caused by manual intervention is reduced, ensuring the accuracy and reliability of feature extraction, and providing a solid data foundation for subsequent anti-fraud tasks.

[0064] In some embodiments, prompt words refer to text or instructions that provide input to the model to guide it to generate specific outputs. Therefore, prompt words are directly related to the quality of the feature extraction code generated by the model, and further affect the quality of subsequent feature extraction. Therefore, good prompt words are particularly important; please refer to Figure 2 , Figure 2 A schematic diagram of a data generation method provided in an embodiment of the present disclosure includes:

[0065] Step 201: input the target prompt word into a training language model to generate a first training feature engineering code; wherein the first training feature engineering code has the same format as the expected feature engineering code.

[0066] See also Figure 3 , Figure 3 The following is a flow chart of a method for generating a prompt word provided in an embodiment of the present application. Figure 3 As shown, when constructing a target prompt word, the user inputs a description of the data set and the required features. The target prompt word includes the context information and semantic information of the data set to be analyzed, the target feature name and the annotation of the target feature and at least one of the expected feature engineering codes.

[0067] Specifically, user-generated dataset descriptions that contain contextual and semantic information about the dataset; feature names that provide contextual information to LLMs and allow LLMs to generate code based on feature names; templates for expected code generation and explanations to guide LLMs in generating feature engineering code and providing explanations. In addition, the prompts include a series of instructions for intermediate reasoning steps that are very effective in improving the quality of LLMs' responses. The entire prompt is designed to provide LLMs with sufficient information and guidance to ensure that the generated feature engineering code and explanations can improve predictive performance and are interpretable.

[0068] The constructed target prompt word is input into the large model (i.e., the training language model), and the training language model generates the training feature engineering code in the same format as the expected feature engineering code in the target prompt word; in some embodiments, code comments can also be generated when generating the training feature engineering code.

[0069] Step 202: Analyze the training data set according to the first training feature engineering code to obtain a first training target feature.

[0070] Please continue reading Figure 3 , the generated training feature engineering code is executed by a code executor to extract features from the training data set; in some embodiments, the training data set may be historical user data or real-time user data, which is not limited in the embodiments of the present application.

[0071] Step 203: Determine, in the first training target features, a second training target feature having a positive performance improvement.

[0072] Please continue reading Figure 3 , the first training target feature is cross-validated through the prediction model. When determining whether the first training feature has a positive effect, it can be determined by calculating the loss function and other methods, such as determining whether the model recognition accuracy is improved and whether the model recall rate is increased. Whether the model F1 score is improved and whether the loss function is reduced, etc., can be set according to actual needs in actual applications, and the embodiments of this application do not limit this.

[0073] In some possible implementations, the combination and interaction effects between features are considered to check whether the feature combination can provide better prediction ability than a single feature, and the interaction between features is analyzed, such as using polynomial features or feature crossing, and using cross-validation techniques to evaluate the performance of feature sets on multiple data subsets.

[0074] Based on the first training target feature, a second training target feature that can bring positive performance improvement is determined.

[0075] Step 204, reconfigure the target prompt word, and train according to the preset target prompt word after the reconfiguration, until the training times reach a preset times threshold, then stop the training; wherein the preset configuration prompt word after the reconfiguration includes the feature name of the second training target feature.

[0076] Please continue reading Figure 3 ,like Figure 3 As shown, based on the results of the first round of training, a performance evaluation is performed to determine the target features for the second training. The target features for the second training are features that are identified in the first round of training as having a significant positive impact on the model performance.

[0077] The feature name of the second training target feature is added to the target prompt word, and the features that have a negative impact on the model performance or have no performance impact are deleted, and the target feature name is re-determined.

[0078] In some embodiments, the weight or priority of the prompt words may be adjusted according to the results of the first round of training, and the relationship between features, such as dependency or mutual exclusivity, may be considered and configured accordingly in the prompt words.

[0079] After the secondary configuration is completed, a new round of training is started using the preset target prompt words after the secondary configuration, and training iterations are performed until the preset training number threshold is reached:

[0080] In each iteration, the model learns the patterns in the dataset based on the new prompt word configuration and monitors the performance of the model after each iteration, including indicators such as loss function value and accuracy.

[0081] In some embodiments, if the model performance does not improve significantly in several consecutive iterations, training can be stopped early to prevent overfitting.

[0082] In some embodiments, step 102 further includes:

[0083] The trained target prompt word is input into the preset language model, and a feature engineering code for extracting the target feature is generated according to the preset language model.

[0084] The target features contained in the trained target prompt words are features that have a positive effect on the subsequent anti-fraud model predictions. Therefore, they can provide precise guidance for the preset language model, so that the generated code can more accurately capture the key attributes in the data, thereby improving the performance and predictive capabilities of subsequent anti-fraud identification.

[0085] In some embodiments, step 102 further includes:

[0086] The preset language model generates an annotation for the target feature; wherein the annotation includes at least one of a feature extraction method and an applicable scenario of the target feature.

[0087] The preset language model provides detailed annotations for the target features, which include the purpose and extraction method of each target feature, as well as the applicability of the target feature in the actual application scenario. For example, data cleaning is to ensure data quality, feature selection is to extract the most relevant information for the task, feature conversion and standardization are to improve the training efficiency and predictive ability of the model, and feature combination is to explore the interaction between features. It should be noted that this narrative method is only an exemplary description, and the embodiments of the present application are not limited to this.

[0088] The account data collected by the bank is mostly structured tabular data, which is easy for the big model to understand and process. The specific plan is as follows:

[0089] 1. The user provides a dataset description and contextual information as input, including data type, missing value ratio, and a random sample of the dataset.

[0090] 2. Build a prompt that contains instructions to the LLMs, guides the LLMs to generate useful feature engineering code, and provides explanations for the generated features.

[0091] 3. Perform multiple feature modifications and verifications. Each time, LLMs generates code and then executes it on the current training set and verification set to obtain the transformed dataset.

[0092] 4. The generated code will modify the dataset, including creating meaningful features, removing unnecessary features, etc.

[0093] 5. Evaluate the performance improvement of the generated features on downstream prediction tasks and retain the features that improve performance.

[0094] LLMs are used to automatically generate feature engineering code and provide explanations for the generated features, making automated feature engineering more transparent and explainable. Further, large language models (LLMs) play the role of generating feature engineering code. LLMs are used to automatically generate feature engineering code, which is used to create new semantically meaningful features to improve the performance of downstream prediction tasks. LLMs understand and reason about dataset descriptions and contextual information to generate feature engineering code that can improve prediction performance.

[0095] The design of Prompt includes the following:

[0096] 1. User-generated dataset descriptions that contain contextual and semantic information about the dataset.

[0097] 2. Feature names, which are used to provide contextual information to LLMs and allow LLMs to generate code based on feature names.

[0098] 3. Templates for expected code generation and explanations to guide LLMs in generating feature engineering code and providing explanations.

[0099] In addition, Prompt includes a series of instructions for intermediate reasoning steps, which are very effective in improving the response quality of LLMs. The entire Prompt is designed to provide LLMs with sufficient information and guidance to ensure that the generated feature engineering code and explanations can improve predictive performance and are interpretable.

[0100] Corresponding to the above data generation method, the present invention also provides a data generation device. Since the device embodiment of the present invention corresponds to the above method embodiment, details not disclosed in the device embodiment can be referred to the above method embodiment, and will not be repeated in the present invention.

[0101] Figure 4 A schematic diagram of a data generation device provided in an embodiment of the present disclosure is shown in FIG. Figure 4 As shown, including:

[0102] A construction unit 31 is used to construct a target prompt word; wherein the target prompt word includes a description of the data set to be analyzed and a target feature name;

[0103] A generating unit 32, configured to input the target prompt word into a preset language model, and generate a feature engineering code for extracting the target feature according to the preset language model;

[0104] The analyzing unit 33 is used to analyze the data set to be analyzed according to the feature engineering code to obtain the target feature.

[0105] The data generation device provided by the present disclosure has a main technical solution including: constructing a target prompt word; wherein the target prompt word includes a description of the data set to be analyzed and a target feature name; inputting the target prompt word into a preset language model, and generating a feature engineering code for extracting the target feature according to the preset language model; analyzing the data set to be analyzed according to the feature engineering code to obtain the target feature. Compared with the related art, the embodiment of the present application automatically generates feature engineering code through the preset language model by inputting the preconfigured prompt word including the data set description and the target feature name into a preset language model, and then processes the data set to be analyzed to obtain the target feature, thereby updating the feature library and providing data support for subsequent anti-fraud tasks. By automatically generating feature engineering code, the tedious process of traditional manual coding is avoided, the speed and efficiency of feature engineering are improved, and the possibility of errors caused by human intervention is reduced, ensuring the accuracy and reliability of feature extraction, and providing a solid data foundation for subsequent anti-fraud tasks.

[0106] Furthermore, in a possible implementation of the embodiment of the present disclosure, the target prompt word includes at least one of the context information and semantic information of the data set to be analyzed, the target feature name and the annotation of the target feature and the expected feature engineering code.

[0107] Furthermore, in a possible implementation of the embodiment of the present disclosure, as Figure 5 As shown, the construction unit 31 also includes:

[0108] An input module 311 is used to input the target prompt word into a training language model to generate a first training feature engineering code; wherein the first training feature engineering code has the same format as the expected feature engineering code;

[0109] An analysis module 312 is used to analyze the training data set according to the first training feature engineering code to obtain a first training target feature;

[0110] A determination module 313, configured to determine a second training target feature having a positive performance improvement in the first training target feature;

[0111] The construction module 314 is used to reconfigure the target prompt word and perform training according to the preset target prompt word after the reconfiguration until the training times reach a preset times threshold, and then stop the training; wherein the preset configuration prompt word after the reconfiguration contains the feature name of the second training target feature.

[0112] Furthermore, in a possible implementation of the embodiment of the present disclosure, the generating unit 32 is further configured to:

[0113] The trained target prompt word is input into the preset language model, and a feature engineering code for extracting the target feature is generated according to the preset language model.

[0114] Furthermore, in a possible implementation of the embodiment of the present disclosure, the generating unit 32 further includes:

[0115] The preset language model generates an annotation for the target feature; wherein the annotation includes at least one of a feature extraction method and an applicable scenario of the target feature.

[0116] It should be noted that the above explanation of the method embodiment is also applicable to the device of the embodiment of the present disclosure, and the principle is the same, which is no longer limited in the embodiment of the present disclosure.

[0117] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0118] Figure 6 A schematic block diagram of an example electronic device 400 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0119] like Figure 6 As shown, the device 400 includes a computing unit 401, which can perform various appropriate actions and processes according to a computer program stored in a ROM (Read-Only Memory) 402 or a computer program loaded from a storage unit 408 to a RAM (Random Access Memory) 403. In the RAM 403, various programs and data required for the operation of the device 400 can also be stored. The computing unit 401, the ROM 402, and the RAM 403 are connected to each other via a bus 404. An I / O (Input / Output) interface 405 is also connected to the bus 404.

[0120] A number of components in the device 400 are connected to the I / O interface 405, including: an input unit 406, such as a keyboard, a mouse, etc.; an output unit 407, such as various types of displays, speakers, etc.; a storage unit 408, such as a disk, an optical disk, etc.; and a communication unit 409, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 409 allows the device 400 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0121] The computing unit 401 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, a CPU (Central Processing Unit), a GPU (Graphic Processing Units), various dedicated AI (Artificial Intelligence) computing chips, various computing units running machine learning model algorithms, a DSP (Digital Signal Processor), and any appropriate processor, controller, microcontroller, etc. The computing unit 401 performs the various methods and processes described above, such as a method for generating data. For example, in some embodiments, the method for generating data may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 408. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 400 via the ROM 402 and / or the communication unit 409. When the computer program is loaded into the RAM 403 and executed by the computing unit 401, one or more steps of the method described above may be performed. Alternatively, in other embodiments, the computing unit 401 may be configured to execute the aforementioned data generation method in any other appropriate manner (for example, by means of firmware).

[0122] Various embodiments of the systems and techniques described above herein may be implemented in digital electronic circuit systems, integrated circuit systems, FPGAs (Field Programmable Gate Arrays), ASICs (Application-Specific Integrated Circuits), ASSPs (Application Specific Standard Products), SOCs (System On Chips), CPLDs (Complex Programmable Logic Devices), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include: being implemented in one or more computer programs that may be executed and / or interpreted on a programmable system including at least one programmable processor that may be a special purpose or general purpose programmable processor that may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0123] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0124] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any suitable combination of the foregoing. More specific examples of machine-readable storage media may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a RAM, a ROM, an EPROM (Electrically Programmable Read-Only-Memory) or a flash memory, an optical fiber, a CD-ROM (Compact Dis sc Read-Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0125] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (Cathode-Ray Tube) or LCD (Liquid Crystal Display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball), through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0126] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: LAN (Local Area Network), WAN (Wide Area Network), the Internet, and blockchain networks.

[0127] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship between the client and the server is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services ("Virtual Private Server", or "VPS" for short). The server may also be a server of a distributed system, or a server combined with a blockchain.

[0128] It should be noted that artificial intelligence is a discipline that studies how computers can simulate certain human thought processes and intelligent behaviors (such as learning, reasoning, thinking, planning, etc.), and includes both hardware-level and software-level technologies. Artificial intelligence hardware technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, and big data processing; artificial intelligence software technologies mainly include computer vision technology, speech recognition technology, natural language processing technology, as well as machine learning / deep learning, big data processing technology, knowledge graph technology, and other major directions.

[0129] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0130] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A method for generating data, characterized in that: include: Constructing a target prompt word; wherein the target prompt word includes a description of the data set to be analyzed and a target feature name; Inputting the target prompt word into a preset language model, and generating a feature engineering code for extracting the target feature according to the preset language model; The data set to be analyzed is analyzed according to the feature engineering code to obtain the target feature.

2. The method according to claim 1, characterized in that: The target prompt word includes at least one of the context information and semantic information of the data set to be analyzed, the target feature name, the annotation of the target feature, and the expected feature engineering code.

3. The method according to claim 2, characterized in that The construction target prompt words also include: Inputting the target prompt word into a training language model to generate a first training feature engineering code; wherein the first training feature engineering code has the same format as the expected feature engineering code; Analyze the training data set according to the first training feature engineering code to obtain a first training target feature; Determining a second training target feature having a positive performance improvement in the first training target feature; The target prompt word is reconfigured, and training is performed according to the preset target prompt word after the reconfiguration, until the training times reach a preset times threshold, and then the training is stopped; wherein the preset configuration prompt word after the reconfiguration includes the feature name of the second training target feature.

4. The method according to claim 3, characterized in that The step of inputting the target prompt word into a preset language model and generating a feature engineering code for extracting the target feature according to the preset language model further comprises: The trained target prompt word is input into the preset language model, and a feature engineering code for extracting the target feature is generated according to the preset language model.

5. The method according to claim 2, characterized in that: The step of inputting the target prompt word into a preset language model and generating a feature engineering code for extracting the target feature according to the preset language model further comprises: The preset language model generates an annotation for the target feature; wherein the annotation includes at least one of a feature extraction method and an applicable scenario of the target feature.

6. A data generating device, characterized in that: include: A construction unit, used to construct a target prompt word; wherein the target prompt word includes a description of the data set to be analyzed and a target feature name; A generating unit, configured to input the target prompt word into a preset language model, and generate a feature engineering code for extracting the target feature according to the preset language model; An analysis unit is used to analyze the data set to be analyzed according to the feature engineering code to obtain the target feature.

7. The device according to claim 6, characterized in that The target prompt word includes at least one of the context information and semantic information of the data set to be analyzed, the target feature name, the annotation of the target feature, and the expected feature engineering code.

8. An electronic device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 5.

9. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-5.

10. A computer program product, characterized in that The invention comprises a computer program which, when executed by a processor, implements the method according to any one of claims 1 to 5.

Images