Data processing system and method, readable medium and electronic equipment

Abstract

The invention relates to a data processing system and method, a readable medium and electronic equipment, and aims to form a data utilization loop, improve the data utilization rate, continuously optimize a model based on user feedback, and improve the performance of the model. The system comprises a prompt generation unit used for acquiring user input data and generating target prompt information according to the user input data; the model processing unit is used for inputting the target prompt information into a target data processing model to obtain a data processing result; the training data generation unit is used for obtaining user feedback information aiming at the data processing result and generating first target training data according to the target prompt information, the data processing result and the user feedback information, and the first target training data is used for carrying out optimization training on the target data processing model.

Description

Technical Field

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

[0002] A large language model (LLM) refers to a deep learning model trained using a large amount of text data. It can generate natural language text or understand the meaning of language text and handle a variety of natural language tasks such as text classification, question and answer, and dialogue.

[0003] In the related art, during the model training process, a large amount of data is needed to train the large language model, the data utilization rate is low, and a model that performs well during the model training process may not perform well during the model application process. Summary of the invention

[0004] This summary is provided to introduce concepts in a brief form that will be described in detail in the detailed description below. This summary is not intended to identify key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0005] In a first aspect, the present disclosure provides a data processing system, the system comprising:

[0006] A prompt generating unit, used for acquiring user input data and generating target prompt information according to the user input data;

[0007] A model processing unit, used for inputting the target prompt information into a target data processing model to obtain a data processing result;

[0008] A training data generating unit is used to obtain user feedback information for the data processing result, and generate first target training data according to the target prompt information, the data processing result and the user feedback information, wherein the first target training data is used to optimize the training of the target data processing model.

[0009] In a second aspect, the present disclosure provides a data processing method, applied to a data processing system, the method comprising:

[0010] Obtaining user input data, and generating target prompt information according to the user input data;

[0011] Inputting the target prompt information into a target data processing model to obtain a data processing result;

[0012] User feedback information for the data processing result is obtained, and first target training data is generated according to the target prompt information, the data processing result and the user feedback information, wherein the first target training data is used to optimize the training of the target data processing model.

[0013] In a third aspect, the present disclosure provides a computer-readable medium having a computer program stored thereon, which, when executed by a processing device, implements the steps performed by any unit in the data processing system described in the first aspect or the steps of the data processing method described in the second aspect.

[0014] In a fourth aspect, the present disclosure provides an electronic device, including:

[0015] a storage device having a computer program stored thereon;

[0016] A processing device is used to execute the computer program in the storage device to implement the steps performed by any unit in the data processing system described in the first aspect or the steps of the data processing method described in the second aspect.

[0017] Through the above technical solution, the prompt generation unit obtains user input data, and generates target prompt information based on the user input data, and then the target prompt information is input into the target data processing model through the model processing unit to obtain the data processing result, and then the training data generation unit obtains user feedback information on the data processing result, and generates the first target training data based on the target prompt information, the data processing result and the user feedback information. The system uses the data in the model application process and the user's feedback information on the data processing result to optimize the data processing model in real time. On the one hand, the optimized data processing model can be used to continue to process data, and new training data can be obtained to continue to optimize the data processing model, and so on, forming a data utilization loop to improve data utilization. On the other hand, the model can be continuously optimized based on user feedback during the model application process to improve model performance.

[0018] Other features and advantages of the present disclosure will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and the originals and elements are not necessarily drawn to scale. In the drawings:

[0020] Figure 1 is a block diagram of a data processing system provided according to an exemplary embodiment;

[0021] Figure 2 is a block diagram of a data processing system provided according to an exemplary embodiment;

[0022] Figure 3 is a block diagram of a data processing system provided according to an exemplary embodiment;

[0023] Figure 4 is a block diagram of a data processing system provided according to an exemplary embodiment;

[0024] Figure 5 is a block diagram of a data processing system provided according to an exemplary embodiment;

[0025] Figure 6 is a flowchart of a data processing method provided according to an exemplary embodiment;

[0026] Figure 7 is a block diagram of an electronic device provided according to an exemplary embodiment. DETAILED DESCRIPTION

[0027] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein, which are instead provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0028] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0029] The term "including" and its variations used herein are open inclusions, i.e., "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.

[0030] It should be noted that the concepts such as "first" and "second" mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0031] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0032] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0033] It is understandable that before using the technical solutions disclosed in the embodiments of the present disclosure, the types, scope of use, usage scenarios, etc. of the personal information involved in the present disclosure should be informed to the user and the user's authorization should be obtained in an appropriate manner in accordance with relevant laws and regulations.

[0034] For example, in response to receiving an active request from a user, a prompt message is sent to the user to clearly prompt the user that the operation requested to be performed will require obtaining and using the user's personal information. Thus, the user can autonomously choose whether to provide personal information to software or hardware such as an electronic device, application, server, or storage medium that performs the operation of the technical solution of the present disclosure according to the prompt message.

[0035] As an optional but non-limiting implementation, in response to receiving an active request from the user, the prompt information may be sent to the user in the form of a pop-up window, in which the prompt information may be presented in text form. In addition, the pop-up window may also carry a selection control for the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0036] It is understandable that the above notification and the process of obtaining user authorization are merely illustrative and do not constitute a limitation on the implementation of the present disclosure. Other methods that meet the relevant laws and regulations may also be applied to the implementation of the present disclosure.

[0037] At the same time, it is understandable that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and relevant provisions.

[0038] Large language models are highly dependent on various data. In related technologies, data is mainly used in the model training process, and the data utilization rate is low. Moreover, a model that performs well in the model training process may not perform well in the model application process. How to efficiently use various data while ensuring data real-time has become an urgent problem to be solved.

[0039] In view of this, the present disclosure provides a data processing system, method, readable medium and electronic device to solve the above technical problems.

[0040] The embodiments of the present disclosure are further explained and illustrated below in conjunction with the accompanying drawings.

[0041] Figure 1 is a block diagram of a data processing system according to an exemplary embodiment of the present disclosure, referring to Figure 1 The data processing system 10 includes a prompt generating unit 101, a model processing unit 102 and a training data generating unit 103.

[0042] The prompt generation unit 101 is used to obtain user input data and generate target prompt information according to the user input data. The model processing unit 102 is used to input the target prompt information into the target data processing model to obtain the data processing result. The training data generation unit 103 is used to obtain user feedback information on the data processing result and generate first target training data according to the target prompt information, the data processing result and the user feedback information. The first target training data is used to optimize the target data processing model.

[0043] By adopting the above-mentioned data processing system, the data in the model application process and the user's feedback information on the data processing results can be used to optimize the data processing model in real time. On the one hand, the optimized data processing model can be used to continue processing data, and new training data can be obtained to continue to optimize the data processing model, and so on, forming a data utilization loop and improving data utilization. On the other hand, the model can be continuously optimized based on user feedback during the model application process to improve model performance.

[0044] It should be understood that user feedback information can be positive feedback information or negative feedback information. Positive feedback information indicates that the user is satisfied with the data processing result, while negative feedback information indicates that the user is dissatisfied with the data processing result. Of course, user feedback information can also be divided into fine-grained levels according to the user's satisfaction level, and the present disclosure does not limit this.

[0045] Furthermore, user feedback information can be obtained through direct feedback and indirect feedback. For example, the data processing system is a question-and-answer system, which is a system that outputs corresponding answers according to the questions input by the user. For example, after outputting the answer (data processing result), the user feedback options can be displayed on the operation interface of the question-and-answer system, such as displaying satisfactory and unsatisfactory options for the user to choose, so that the corresponding user feedback information can be obtained after the user selects the corresponding option, that is, the user feedback information is obtained through direct feedback. The corresponding user feedback information can also be obtained according to the user's operation, such as the user's stay time, whether to continue asking questions, etc. For example, if the user asks the same question again, it means that the user is not satisfied with the original output answer. If the user enters other questions, it means that the user is satisfied with the original output answer, that is, the user feedback information is obtained through indirect feedback, and the present disclosure does not limit this.

[0046] In possible ways, refer to Figure 2 The data processing system 10 also includes a data collection unit 104, which can be used to acquire multi-source data in real time and perform data verification on the multi-source data to obtain target multi-source data. The data verification includes at least one of performing a horizontal comparison of multi-source data and comparing the multi-source data with historical multi-source data.

[0047] For example, multi-source data is obtained through a data collection unit to build a knowledge base (Knowledge Base), and the multi-source data includes data from different channels such as internal data and external data of the data processing system, such as facts, public opinion, etc., and the present disclosure does not limit this. In order to ensure the timeliness and factuality of the data, the latest data can be obtained from various channels in real time, or the data can be obtained once at a preset time interval, such as obtaining data once every hour, and the present disclosure does not limit this. Since the quality of the multi-source data obtained varies, such as there may be false or erroneous data, the obtained data can be compared with historical data, or the data from multiple channels can be compared horizontally across multiple channels, etc. The present disclosure does not limit this, so that the correctness of the data can be automatically verified.

[0048] In a possible manner, the data collection unit is further used to perform filtering processing and / or classification processing on the target multi-source data through the target data processing model to obtain the target multi-source data after model processing.

[0049] For example, continue to refer to Figure 2The target data processing model has data processing capabilities, so the data processing model can be used to analyze and count multi-source data, filter out false and invalid data, obtain high-quality data, and classify the data, etc., which is not limited by the present disclosure. Further, the target multi-source data processed by the model is stored, that is, a knowledge base is constructed.

[0050] In this way, taking the data processing system as a question-answering system as an example, it is possible to realize fully automatic collection and verification of data to build a knowledge base. The target data processing model can output answers based on the knowledge base, and then generate training data based on the answers and user feedback information to optimize the target data processing model. The optimized and trained target data processing model can also be used to process multi-source data, realizing a data utilization loop and improving data utilization efficiency and model processing performance.

[0051] It is worth noting that the data in the knowledge base can also be used to generate prompt information (Prompt), and can also be used to verify data processing results.

[0052] In a possible manner, the prompt generating unit may be used to generate target prompt information according to target multi-source data and user input data, where the target multi-source data includes internal data and / or external data of the data processing system.

[0053] For example, continue to refer to Figure 2 Continuing with the example of the data processing system being a question-answering system, the prompt generation unit can search for relevant data in the knowledge base based on the questions input by the user. For example, when the user asks a question related to a certain field, the relevant knowledge in the field can be searched in the knowledge base, and prompt information can be generated in combination with the question input by the user, so that the data processing model can better understand the user's question and output an answer, thereby improving data utilization efficiency and model processing performance.

[0054] In a possible manner, the prompt generation unit can be used to generate target prompt information based on user input data and a target prompt template, where the target prompt template includes at least one of a first prompt template generated by a template generation model and a manually configured second prompt template; or the user input data is input into the prompt generation model to obtain the target prompt information.

[0055] For example, the target prompt information can be generated according to the user input data and the target prompt template. The target prompt template can be a first prompt template generated by a template generation model, or a second prompt template manually configured, which is not limited in the present disclosure. Alternatively, the user input data is input into a prompt generation model, such as a text generation model, to obtain the target prompt information.

[0056] It should be understood that whether the target prompt information is generated based on the target prompt template or the target prompt information is generated based on the prompt generation model, it meets the input requirements of the data processing model, such as format requirements, etc., so that the data processing model can recognize the target prompt information, thereby improving the processing efficiency of the data processing model.

[0057] In possible ways, refer to Figure 3 The data processing system 10 further includes a data result verification unit 105; the data result verification unit 105 is used to verify the correctness of the data processing result according to the target multi-source data, and determine the data processing result that passes the correctness verification as the target data processing result, and the target multi-source data includes the internal data and / or external data of the data processing system. The training data generation unit 103 is also used to obtain the target user feedback information for the target data processing result, and generate the first target training data according to the target prompt information, the target data processing result and the target user feedback information.

[0058] For example, continuing to take the data processing system as a question-answering system, the answer output by the target data processing model may be wrong, so refer to Figure 3 , the data processing result can be verified for correctness based on the target multi-source data through the data result verification unit. For example, if a user asks a question related to a certain field, the data result verification unit can verify the correctness of the answer in combination with the data in the field. If the correctness verification is passed, the answer will be output and displayed on the operation interface of the question-answering system. If it fails, it will be fed back to the model processing unit so that the target data processing model can re-process the data, thereby improving the correctness of the data processing.

[0059] Correspondingly, the training data generation unit obtains the target data processing results that have passed the correctness verification and the target user feedback information for the target data processing results, and generates the first target training data based on the target prompt information, the target data processing results and the target user feedback information, which can improve the correctness of the training data and thereby improve the performance of the target data processing model.

[0060] In the process of data processing by the above data processing system, the latest data can be obtained from various channels in real time, and multi-source data can be verified and modeled to build a knowledge base. The data in the knowledge base can assist the prompt generation unit in generating prompt information, assist the target data processing model in data processing, and verify the correctness of the data processing results. Then, the user's feedback information on the data processing results is obtained to optimize the target data processing model for training. The optimized and trained target data processing model can continue to process multi-source data to form a data utilization loop. While ensuring the timeliness and correctness of the data, the data utilization rate is improved. On the other hand, the model is continuously optimized based on user feedback to ensure that the model optimization is in the right direction and improve the model performance.

[0061] It is worth noting that the data processing system can be divided into two parts: front-end data processing and back-end optimization. The process of front-end data processing has been described in detail in the above embodiments, and the process of back-end optimization is described in detail below.

[0062] In a possible manner, the training data generating unit is further configured to generate second target training data according to the target prompt information and the user feedback information. Figure 4 The data processing system 10 also includes a prompt training unit 106, which can be used to: optimize the template generation model according to the second target training data, and screen the target prompt template according to the second target training data; or optimize the prompt generation model according to the second target training data.

[0063] For example, the prompt training unit can generate second target training data according to the target prompt information and the user feedback information, and then optimize the template generation model according to the second target training data. For example, a positive sample is constructed according to the positive feedback information and the target prompt information corresponding to the positive feedback information, and a negative sample is constructed according to the negative feedback information and the target prompt information corresponding to the negative feedback information, and then model adversarial training is performed to optimize the template generation model. The present disclosure does not limit the sample construction method and model training process of the template generation model, as long as the model optimization is ensured to be carried out in the right direction.

[0064] It is worth noting that, whether it is the first prompt template generated by the template generation model or the manually configured second prompt template, multiple prompt templates can be generated in batches. Therefore, the prompt templates can be screened according to user feedback information, and the prompt templates that generate the target prompt information corresponding to the positive feedback information are retained, thereby optimizing the process of generating prompt information.

[0065] Alternatively, second target training data can be generated based on the target prompt information and user feedback information, and then the prompt generation model can be optimized and trained based on the second target training data. Similarly, the present disclosure does not limit the sample construction method and model training process of the prompt generation model, as long as the model optimization is carried out in the right direction.

[0066] It should be understood that the training data generating unit may also obtain the target user feedback information for the target data processing result, and generate the second target training data according to the target prompt information and the target user feedback information, and the present disclosure does not limit this.

[0067] Therefore, the user feedback information can be used to optimize the prompt generation process, realize the data utilization loop of the prompt generation process, and improve the data utilization rate.

[0068] In a possible manner, the number of first target training data is multiple, and the user feedback information corresponding to the multiple first target training data includes positive feedback information and negative feedback information. The training data generation unit is used to generate the first training data according to the positive feedback information and the first prompt information corresponding to the positive feedback information and the first data processing result, and to generate the second training data according to the negative feedback information and the second prompt information corresponding to the negative feedback information and the second data processing result; wherein the first training data is used to perform initial model training on the target data processing model to obtain the initially optimized target data processing model, and the first training data and the second training data are used together to perform secondary training on the initially optimized target data processing model to obtain the secondary optimized target data processing model.

[0069] For example, taking the target data processing model as a large language model, the target data processing model can be optimized and trained using the SFT+RLHF method, where SFT (Supervised Fine-tuning Model) represents a supervised fine-tuning model, and RLHF (Reinforcement Learning from Human Feedback) represents reinforcement learning based on human feedback. In the SFT model optimization process, the first training data is used to optimize the model, and in the RLHF model optimization process, the first training data and the second training data are used to jointly optimize the model to ensure that the model optimization is carried out in the right direction and improve the model processing performance.

[0070] In a possible manner, the system also includes a model training unit, which can be used to perform initial model training on multiple data processing models corresponding to the target data processing model through first training data to obtain multiple initial optimized data processing models; for each initial optimized data processing model, the initial optimized data processing model is secondarily trained through the first training data and the second training data to obtain a secondary optimized data processing model; a target update model is determined based on multiple secondary optimized data processing models corresponding to the multiple initial optimized data processing models, and the target data processing model is updated based on the target update model to obtain a new target data processing model.

[0071] For example, refer to Figure 5 The data processing system also includes a model training unit. Taking the target data processing model as a large language model as an example, the model training is divided into two parts: SFT and RLHF. It should be understood that multiple data processing models can be optimized in batches to select the data processing model with the best performance. First, in the SFT model optimization process, the first training data is used to optimize the model to obtain the initial optimized data processing model, and then in the RLHF model optimization process, the first training data and the second training data are used to jointly optimize the model to obtain a secondary optimized data processing model, so as to ensure that the model optimization is carried out in the right direction and improve the model processing performance.

[0072] In a possible manner, the model training unit is used to: for each secondary optimized data processing model, input preset verification data into the secondary optimized data processing model to obtain a first output result, and determine a first verification score corresponding to the secondary optimized data processing model based on the preset verification result corresponding to the preset verification data and the first output result; and determine the secondary optimized data processing model with the highest first verification score among multiple secondary optimized data processing models as the target update model.

[0073] For example, sample data for verification, i.e., preset verification data, can be pre-configured, and the trained data processing model can be verified by the preset verification data. By comparing the model output results with the preset verification results corresponding to the preset verification data, a score can be given based on the degree of closeness between the model output results and the preset verification results. The higher the degree of closeness, the higher the score. The data processing model with the highest score is thus selected as the target update model, and the target data processing model is updated to improve the model data processing performance.

[0074] In a possible manner, the model training unit is used to: for each secondary optimized data processing model, input the first prompt information corresponding to the positive feedback information into the secondary optimized data processing model to obtain a second output result, and determine the second verification score corresponding to the secondary optimized data processing model based on the first data processing result and the second output result corresponding to the positive feedback information; and determine the secondary optimized data processing model with the highest second verification score among multiple secondary optimized data processing models as the target update model.

[0075] For example, the trained data processing model is verified by the first data processing result corresponding to the positive feedback information. By comparing the model output result with the first data processing result, a score can be given according to the degree of closeness between the model output result and the first data processing result. The higher the degree of closeness, the higher the score. The data processing model with the highest score is thus selected as the target update model, and the target data processing model is updated to improve the model data processing performance.

[0076] It is worth noting that the above-mentioned process of screening the trained data processing model can be set according to the needs, for example, different answer verification dimensions can be set for scoring, and the present disclosure does not limit this. By automatically obtaining different data processing models for verification, and automatically selecting the optimal model through the answer score, and the actual score can be fed back to the target data processing model, the target data processing model is assisted to be optimized according to the optimal model.

[0077] It is worth noting that the optimization of the prompt information generation process and the optimization of the data processing model can be performed simultaneously or separately, and the present disclosure does not impose any restrictions on this.

[0078] By adopting the above scheme, by obtaining user feedback information on the data processing results, on the one hand, it is integrated into training data, based on which the design and generation process of Prompt is optimized, and on the other hand, user feedback, prompt information and data processing results are integrated into training data, and the data processing model is optimized and trained by SFT+RLHF. The data processing system can support the data processing process and optimization process in the model application process, making it a system that can be iterated for a long time, and can form a channel for user feedback data, multi-source data and model application. Each link will cross-use and verify data, which is equivalent to connecting the various units of the data processing system in series with data as the center, so that the data flow channel becomes systematic, and each unit of the system uses the data, and finally optimizes the data processing system, thereby improving the utilization rate of data and the performance of model processing.

[0079] It should be understood that the data processing model can be a large language model, such as a text generation model, or it can be a picture generation model, a picture-text mixed generation model, and the like, and the present disclosure does not limit this.

[0080] Based on the same inventive concept, the present disclosure provides a data processing method, which is applied to a data processing system, referring to Figure 6 , the method comprising:

[0081] S601: Obtain user input data, and generate target prompt information according to the user input data.

[0082] S602: Input the target prompt information into the target data processing model to obtain the data processing result.

[0083] S603: Obtain user feedback information on the data processing result, and generate first target training data according to the target prompt information, the data processing result and the user feedback information.

[0084] Among them, the first target training data is used to optimize the training of the target data processing model.

[0085] By adopting the above data processing method, the data in the model application process and the user's feedback information on the data processing results can be used to optimize the data processing model in real time. On the one hand, the optimized data processing model can be used to continue processing data, and new training data can be obtained to continue to optimize the data processing model, and so on, forming a data utilization loop and improving data utilization. On the other hand, the model can be continuously optimized based on user feedback during the model application process to improve model performance.

[0086] In a possible manner, generating target prompt information based on user input data may include: generating target prompt information based on user input data and a target prompt template, the target prompt template including at least one of a first prompt template generated by a template generation model and a manually configured second prompt template; or, inputting user input data into the prompt generation model to obtain the target prompt information.

[0087] In a possible manner, the method may also include: generating second target training data based on target prompt information and user feedback information; optimizing and training the template generation model based on the second target training data, and screening the target prompt template based on the second target training data; or optimizing and training the prompt generation model based on the second target training data.

[0088] In a possible manner, generating target prompt information according to user input data may include: generating target prompt information according to target multi-source data and the user input data, wherein the target multi-source data includes internal data and / or external data of the data processing system.

[0089] In a possible manner, the method may further include: verifying the correctness of the data processing result according to the target multi-source data, and determining the target data processing result that passes the correctness verification as the target data processing result, wherein the target multi-source data includes internal data and / or external data of the data processing system. Obtaining target user feedback information for the target data processing result, and generating the first target training data according to the target prompt information, the target data processing result and the target user feedback information.

[0090] In a possible manner, the method may further include: acquiring multi-source data in real time, and performing data verification on the multi-source data to obtain target multi-source data, wherein the data verification includes at least one of performing a horizontal comparison of multi-source data and comparing the multi-source data with historical multi-source data.

[0091] In a possible manner, the method may further include: performing filtering processing and / or classification processing on the target multi-source data through the target data processing model to obtain the target multi-source data after model processing;

[0092] In a possible manner, the number of first target training data is multiple, and the user feedback information corresponding to the multiple first target training data includes positive feedback information and negative feedback information; generating the first target training data according to the target prompt information, the data processing result and the user feedback information may include: generating the first training data according to the positive feedback information and the first prompt information corresponding to the positive feedback information and the first data processing result, and generating the second training data according to the negative feedback information and the second prompt information corresponding to the negative feedback information and the second data processing result; wherein the first training data is used to perform initial model training on the target data processing model to obtain the initially optimized target data processing model, and the first training data and the second training data are used together to perform secondary training on the initially optimized target data processing model to obtain the secondary optimized target data processing model.

[0093] In a possible manner, the method may also include: performing initial model training on multiple data processing models corresponding to the target data processing model through the first training data to obtain multiple initial optimized data processing models; for each initial optimized data processing model, performing secondary training on the initial optimized data processing model through the first training data and the second training data to obtain a secondary optimized data processing model; determining a target update model based on multiple secondary optimized data processing models corresponding to the multiple initial optimized data processing models, and updating the target data processing model based on the target update model to obtain a new target data processing model.

[0094] In a possible manner, determining a target update model based on multiple secondary optimized data processing models corresponding to multiple initial optimized data processing models may include: for each secondary optimized data processing model, inputting preset verification data into the secondary optimized data processing model to obtain a first output result, and determining a first verification score corresponding to the secondary optimized data processing model based on the preset verification result corresponding to the preset verification data and the first output result; and determining the secondary optimized data processing model with the highest first verification score among the multiple secondary optimized data processing models as the target update model.

[0095] In a possible manner, determining a target update model based on multiple secondary optimized data processing models corresponding to multiple initial optimized data processing models may include: for each secondary optimized data processing model, inputting first prompt information corresponding to positive feedback information into the secondary optimized data processing model to obtain a second output result, and determining a second verification score corresponding to the secondary optimized data processing model based on the first data processing result and the second output result corresponding to the positive feedback information; and determining the secondary optimized data processing model with the highest second verification score among the multiple secondary optimized data processing models as the target update model.

[0096] Regarding the data processing method in the above embodiment, the specific manner in which each unit in the data processing system performs operations has been described in detail in the embodiment of the system, and will not be elaborated here.

[0097] Based on the same concept, an embodiment of the present disclosure also provides a computer-readable medium on which a computer program is stored. When the program is executed by a processing device, the steps of the above-mentioned data processing method or the steps executed by any unit of the data processing system are implemented.

[0098] Based on the same concept, an embodiment of the present disclosure further provides an electronic device, including:

[0099] a storage device having a computer program stored thereon;

[0100] A processing device is used to execute the computer program in the storage device to implement the steps of the above-mentioned data processing method or the steps performed by any unit of the data processing system.

[0101] It should be understood that the above data processing method can be implemented by one electronic device, that is, one data processing system corresponds to one electronic device, or it can be implemented by multiple electronic devices, that is, each unit in the data processing system corresponds to one electronic device, and the present disclosure does not limit this.

[0102] Reference below Figure 7, which shows a schematic diagram of the structure of an electronic device 700 suitable for implementing the embodiment of the present disclosure. The electronic device in the embodiment of the present disclosure may include, but is not limited to, mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 7 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0103] like Figure 7 As shown, the electronic device 700 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 701, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 702 or a program loaded from a storage device 708 into a random access memory (RAM) 703. In the RAM 703, various programs and data required for the operation of the electronic device 700 are also stored. The processing device 701, the ROM 702, and the RAM 703 are connected to each other via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0104] Typically, the following devices may be connected to the I / O interface 705: an input device 706 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 707 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 708 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 709. The communication device 709 may allow the electronic device 700 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 7 The electronic device 700 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0105] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device 709, or installed from a storage device 708, or installed from a ROM 702. When the computer program is executed by the processing device 701, the above-mentioned functions defined in the method of the embodiment of the present disclosure are executed.

[0106] It should be noted that the computer-readable medium disclosed above may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0107] In some embodiments, any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol) can be used for communication, and can be interconnected with any form or medium of digital data communication (e.g., communication network). Examples of communication networks include local area networks ("LAN"), wide area networks ("WAN"), internets (e.g., the Internet), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future developed networks.

[0108] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0109] The computer-readable medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device: obtains user input data and generates target prompt information based on the user input data; inputs the target prompt information into a target data processing model to obtain a data processing result; obtains user feedback information on the data processing result, and generates first target training data based on the target prompt information, the data processing result and the user feedback information, and the first target training data is used to optimize the target data processing model.

[0110] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including, but not limited to, object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0111] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0112] The modules involved in the embodiments described in the present disclosure may be implemented by software or hardware, wherein the name of a module does not, in some cases, limit the module itself.

[0113] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0114] 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 equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0115] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present disclosure is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present disclosure (but not limited to) by each other to form a technical solution.

[0116] In addition, although each operation is described in a specific order, this should not be understood as requiring these operations to be performed in the specific order shown or in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be interpreted as limiting the scope of the present disclosure. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.

[0117] Although the subject matter has been described in language specific to structural features and / or method logic actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. On the contrary, the specific features and actions described above are merely example forms of implementing the claims. Regarding the device in the above embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated here.

Claims

1. A data processing system, characterized in that: The system comprises: A prompt generating unit, used for acquiring user input data and generating target prompt information according to the user input data; A model processing unit, used for inputting the target prompt information into a target data processing model to obtain a data processing result; A training data generating unit is used to obtain user feedback information for the data processing result, and generate first target training data according to the target prompt information, the data processing result and the user feedback information, wherein the first target training data is used to optimize the training of the target data processing model.

2. The system according to claim 1, characterized in that The prompt generating unit is used for: Generate the target prompt information according to the user input data and a target prompt template, wherein the target prompt template includes at least one of a first prompt template generated by a template generation model and a second prompt template configured manually; or, The user input data is input into a prompt generation model to obtain the target prompt information.

3. The system according to claim 2, characterized in that The training data generating unit is further used to generate second target training data according to the target prompt information and the user feedback information; The system further comprises a prompt training unit, wherein the prompt training unit is used for: Optimizing and training the template generation model according to the second target training data, and screening the target prompt template according to the second target training data; or, The prompt generation model is optimized and trained according to the second target training data.

4. The system according to claim 1, characterized in that The prompt generating unit is used to generate the target prompt information according to the target multi-source data and the user input data, wherein the target multi-source data includes the internal data and / or external data of the data processing system.

5. The system according to claim 1, characterized in that The system also includes a data result verification unit; The data result verification unit is used to verify the correctness of the data processing result according to the target multi-source data, and determine the data processing result that passes the correctness verification as the target data processing result, wherein the target multi-source data includes internal data and / or external data of the data processing system; The training data generating unit is further used to obtain target user feedback information for the target data processing result, and generate the first target training data according to the target prompt information, the target data processing result and the target user feedback information.

6. The system according to any one of claims 1 to 5, characterized in that: The system further comprises a data collection unit, wherein the data collection unit is configured to: Acquire multi-source data in real time, and perform data verification on the multi-source data to obtain the target multi-source data, wherein the data verification includes at least one of performing a multi-source data source horizontal comparison on the multi-source data and comparing the multi-source data with historical multi-source data.

7. The system according to claim 6, characterized in that The data collection unit is also used for: The target multi-source data is filtered and / or classified by the target data processing model to obtain the target multi-source data after model processing.

8. The system according to any one of claims 1 to 5, characterized in that: The number of the first target training data is multiple, and the user feedback information corresponding to the multiple first target training data includes positive feedback information and negative feedback information; The training data generating unit is configured to generate first training data according to the positive feedback information, the first prompt information corresponding to the positive feedback information, and the first data processing result, and to generate second training data according to the negative feedback information, the second prompt information corresponding to the negative feedback information, and the second data processing result; Among them, the first training data is used to perform initial model training on the target data processing model to obtain an initially optimized target data processing model, and the first training data and the second training data are used together to perform secondary training on the initially optimized target data processing model to obtain a secondary optimized target data processing model.

9. The system according to claim 8, characterized in that The system further comprises a model training unit, wherein the model training unit is used to: Performing initial model training on multiple data processing models corresponding to the target data processing model using the first training data to obtain multiple initial optimized data processing models; For each of the initially optimized data processing models, performing secondary training on the initially optimized data processing model using the first training data and the second training data to obtain a secondary optimized data processing model; A target update model is determined according to the multiple secondary optimized data processing models corresponding to the multiple initially optimized data processing models, and the target data processing model is updated based on the target update model to obtain a new target data processing model.

10. The system according to claim 9, characterized in that The model training unit is used for: For each of the secondary optimized data processing models, inputting preset verification data into the secondary optimized data processing model to obtain a first output result, and determining a first verification score corresponding to the secondary optimized data processing model according to the preset verification result corresponding to the preset verification data and the first output result; The secondary optimized data processing model with the highest first verification score among the multiple secondary optimized data processing models is determined as the target update model.

11. The system according to claim 9, characterized in that The model training unit is used for: For each of the secondary optimized data processing models, the first prompt information corresponding to the positive feedback information is input into the secondary optimized data processing model to obtain a second output result, and a second verification score corresponding to the secondary optimized data processing model is determined according to the first data processing result corresponding to the positive feedback information and the second output result; The secondary optimized data processing model with the highest second verification score among the multiple secondary optimized data processing models is determined as the target update model.

12. A data processing method, characterized in that: Applied to a data processing system, the method comprises: Obtaining user input data, and generating target prompt information according to the user input data; Inputting the target prompt information into a target data processing model to obtain a data processing result; User feedback information for the data processing result is obtained, and first target training data is generated according to the target prompt information, the data processing result and the user feedback information, wherein the first target training data is used to optimize the training of the target data processing model.

13. A computer readable medium having a computer program stored thereon, characterized in that: When the program is executed by a processing device, the program implements the steps performed by any unit in the data processing system according to any one of claims 1 to 11 or the steps of the data processing method according to claim 12.

14. An electronic device, characterized in that: include: a storage device having a computer program stored thereon; A processing device for executing the computer program in the storage device to implement the steps performed by any unit in the data processing system of any one of claims 1 to 11 or the steps of the data processing method of claim 12.

Images