Data processing method and device, storage medium, electronic equipment and computer program product

Abstract

The embodiment of the invention provides a data processing method and device, a storage medium, electronic equipment and a computer program product.The data processing method comprises the steps that in response to a first input operation of a user, finance and tax problem data asked by the user is obtained; performing feature extraction on the finance and taxation problem data to obtain feature data in the finance and taxation problem data; the feature data are processed based on a pre-trained target question and answer model, finance and tax reply data are output, and the target question and answer model is a neural network model obtained through training of a sample data pair composed of historical finance and tax question data samples and corresponding historical finance and tax reply data samples and finance and tax policy data samples; and feeding back the finance and tax reply data. According to the method provided by the embodiment of the invention, the answering efficiency of finance and taxation questions proposed by the user is improved, and the labor cost is reduced.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of computer technology, and in particular, to a data processing method, device, storage medium, electronic device, and computer program product. Background Art

[0002] With the continuous updating of fiscal and taxation policies and the increasing demand for fiscal and taxation consulting from enterprises, it is particularly important to accurately and efficiently answer fiscal and taxation questions for users. In the existing technology, the answering methods for fiscal and taxation questions usually rely on manual online or offline answers, which leads to low efficiency of answering questions and high labor costs. Summary of the invention

[0003] In view of this, embodiments of the present application provide a data processing method, apparatus, storage medium, electronic device, and computer program product to at least partially solve the above-mentioned problems.

[0004] According to a first aspect of an embodiment of the present application, a data processing method is provided, including: in response to a first input operation of a user, obtaining fiscal and taxation question data asked by the user; performing feature extraction on the fiscal and taxation question data to obtain feature data in the fiscal and taxation question data; processing the feature data based on a pre-trained target question and answer model, and outputting fiscal and taxation response data, wherein the target question and answer model is a neural network model obtained by training sample data pairs consisting of historical fiscal and taxation question data samples and corresponding historical fiscal and taxation response data samples, as well as fiscal and taxation policy data samples; and feeding back the fiscal and taxation response data.

[0005] In some optional embodiments, the training process of the target question and answer model includes: obtaining training data to be processed; performing feature extraction on the training data to be processed to obtain training data; the training data includes sample data pairs consisting of historical fiscal and taxation question data samples and corresponding historical fiscal and taxation response data samples, as well as fiscal and taxation policy data samples; based on the training data, performing supervised training on the preset question and answer model to be trained to obtain the target question and answer model.

[0006] In some optional embodiments, the data processing method of this embodiment also includes: obtaining fiscal and taxation policy data of the current period according to a preset period; if there is a difference between the fiscal and taxation policy data of the current period and the fiscal and taxation policy data of the previous period, constructing an incremental data sample based on the fiscal and taxation policy data of the current period; based on the incremental data sample, performing online update training on the target question and answer model to obtain an updated target question and answer model.

[0007] In some optional embodiments, feature extraction is performed on the training data to be processed to obtain training data, including: deduplication processing is performed on the training data to be processed to obtain preprocessed training data; natural language processing is performed on the preprocessed training data to obtain training data.

[0008] In some optional embodiments, the data processing method of this embodiment also includes: obtaining test data, the test data including sample data pairs consisting of historical fiscal and taxation problem data samples and corresponding historical fiscal and taxation response data samples; based on the target question and answer model, inferring the historical fiscal and taxation problem data samples in the test data to obtain test fiscal and taxation response data; based on the test fiscal and taxation response data and the corresponding historical fiscal and taxation response data samples, determining an evaluation result for the target question and answer model; and optimizing the target question and answer model based on the evaluation result.

[0009] In some optional embodiments, the data processing method of this embodiment also includes: in response to the user's second input operation, obtaining the user's evaluation information on the financial and taxation response data; analyzing the evaluation information and the financial and taxation question data and financial and taxation response data corresponding to the evaluation information to obtain analysis results; and optimizing the target question and answer model based on the analysis results.

[0010] According to a second aspect of an embodiment of the present application, a data processing device is provided, including: an acquisition module, for acquiring, in response to a first input operation of a user, fiscal and taxation question data asked by a user; a preprocessing module, for performing feature extraction on the fiscal and taxation question data to obtain feature data in the fiscal and taxation question data; a model inference module, for processing the feature data based on a pre-trained target question and answer model, and outputting fiscal and taxation response data, wherein the target question and answer model is a neural network model obtained by training sample data pairs consisting of historical fiscal and taxation question data samples and corresponding historical fiscal and taxation response data samples, as well as fiscal and taxation policy data samples; and an interaction module, for feeding back the fiscal and taxation response data.

[0011] According to a third aspect of an embodiment of the present application, a computer storage medium is provided, on which a computer program is stored, and when the program is executed by a processor, the method of the first aspect is implemented.

[0012] According to the fourth aspect of an embodiment of the present application, there is provided an electronic device, comprising a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; the memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform an operation corresponding to the method of the first aspect.

[0013] According to a fifth aspect of an embodiment of the present application, a computer program product is provided, comprising computer instructions, which instruct a computing device to perform operations corresponding to the method of the first aspect.

[0014] The embodiments of the present application provide a data processing method, device, storage medium, electronic device and computer program product, wherein the data processing method includes: in response to a first input operation of a user, obtaining the fiscal and taxation question data asked by the user; extracting features from the fiscal and taxation question data to obtain feature data in the fiscal and taxation question data; processing the feature data based on a pre-trained target question-answering model to output fiscal and taxation response data, wherein the target question-answering model is a neural network model trained through sample data pairs consisting of historical fiscal and taxation question data samples and corresponding historical fiscal and taxation response data samples, as well as fiscal and taxation policy data samples; and feeding back the fiscal and taxation response data. The embodiments of the present application extract features from the fiscal and taxation question data asked by the user to obtain feature data in the fiscal and taxation question data; and then use the pre-trained target question-answering model to process the feature data to output fiscal and taxation response data without manual answers, thereby improving the efficiency of answering fiscal and taxation questions raised by users and reducing labor costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the embodiments of the present application. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0016] Figure 1 is a flowchart of a data processing method according to an exemplary embodiment of the present application;

[0017] Figure 2 A schematic diagram of an interactive interface of a data processing method for an exemplary embodiment of the present application;

[0018] Figure 3 is a structural block diagram of a data processing device according to an exemplary embodiment of the present application;

[0019] Figure 4 It is a schematic structural diagram of an electronic device according to an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0020] In order to enable those skilled in the art to better understand the technical solutions in the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments in the embodiments of the present application should fall within the scope of protection of the embodiments of the present application.

[0021] Reference Figure 1, shows a flow chart of steps of a data processing method according to an exemplary embodiment of the present application. As shown in the figure, this embodiment mainly includes the following steps:

[0022] S102. In response to a first input operation by the user, obtain financial and taxation question data asked by the user.

[0023] For example, refer to Figure 2 , an interactive interface can be provided to the user, the interactive interface includes a first input area, the first input operation can be a question input operation of the user in the first input area, and the first input operation of the user in the interactive interface can be detected to obtain the financial and taxation question data asked by the user, and the financial and taxation question data can usually be text data.

[0024] S104. Extract features from the financial and taxation problem data to obtain feature data from the financial and taxation problem data.

[0025] For example, the financial and taxation problem data can be pre-processed by word segmentation, part-of-speech tagging, etc. to extract characteristic data from the financial and taxation problem data. Among them, word segmentation is to divide the continuous natural language text in the financial and taxation problem data into individual words according to certain rules. Part-of-speech tagging is to determine the part of speech of each word obtained after word segmentation, such as noun, verb, adjective, etc.

[0026] S106. Process the feature data based on the pre-trained target question-answering model and output financial and tax response data.

[0027] Exemplarily, the feature data can be input into a pre-trained target question-answering model, which can process the feature data and output corresponding financial and tax response data.

[0028] Among them, the target question-answering model is a neural network model trained through sample data pairs consisting of historical fiscal and taxation question data samples and corresponding historical fiscal and taxation response data samples, as well as fiscal and taxation policy data samples.

[0029] For example, data samples can be collected from various channels such as taxation departments, taxation consulting agencies, and historical question and answer records. Data samples may include taxation policy documents of various regions, official explanations of taxation departments, historical question and answer records, and knowledge bases of taxation experts. Historical question and answer records may include historical taxation question data samples and corresponding historical taxation answer data samples. Sample data pairs consisting of historical taxation question data samples and corresponding historical taxation answer data samples can be generated based on taxation policy documents, official explanations of taxation departments, and knowledge bases of taxation experts.

[0030] In some optional embodiments, the training process of the target question and answer model includes: obtaining training data to be processed; performing feature extraction on the training data to be processed to obtain training data; the training data includes sample data pairs consisting of historical fiscal and taxation question data samples and corresponding historical fiscal and taxation response data samples, as well as fiscal and taxation policy data samples; based on the training data, performing supervised training on the preset question and answer model to be trained to obtain the target question and answer model.

[0031] Exemplarily, the training data to be processed may include data samples collected from channels such as taxation departments in various regions, taxation consulting agencies, and historical question and answer records. The data samples may include fiscal and taxation policy documents of various regions, official explanations of taxation departments, historical question and answer records, and knowledge bases of fiscal and taxation experts. The training data to be processed may be processed by data cleaning, classification, and labeling, for example, removing duplicate and invalid data; converting fiscal and taxation policy documents, official explanations of taxation departments, and knowledge bases of fiscal and taxation experts into structured data that meets the training requirements of the preset question and answer model, that is, generating fiscal and taxation policy data samples; historical question and answer records may include historical fiscal and taxation question data samples and corresponding historical fiscal and taxation response data samples, and performing intent recognition and answer extraction on historical question and answer records, and sample data pairs consisting of historical fiscal and taxation question data samples and corresponding historical fiscal and taxation response data samples may be formed.

[0032] Pre-construct a preset question-answering model to be trained. For example, the preset question-answering model may include an input layer, an encoding layer, and a decoding layer, wherein the encoding layer may adopt the encoder part of the Transformer model or other encoding methods, which is not limited in this embodiment. A suitable optimizer and loss function may be selected according to actual needs, wherein the optimizer may be an Adam optimization algorithm, an SGD (stochastic gradient descent) optimization algorithm, etc., and the loss function may be a cross entropy loss function, a mean square error loss function, etc., which is not limited in this embodiment. In addition, during the training process, the learning rate of the model may be adjusted to prevent the model from falling into a local optimal solution; and L1 / L2 regularization, Dropout and other technologies may be used to prevent the model from overfitting and improve the generalization ability of the model. Dropout refers to the process of training a deep learning network, in which a neural network unit is temporarily discarded from the network according to a certain probability.

[0033] In this embodiment, training data is obtained by performing feature extraction on the training data to be processed; then, based on the training data, supervised training is performed on the preset question-answering model to be trained to obtain a target question-answering model, which can reduce the noise interference of the training data to improve the accuracy of the target question-answering model.

[0034] In some optional embodiments, feature extraction is performed on the training data to be processed to obtain training data, including: deduplication processing is performed on the training data to be processed to obtain preprocessed training data; natural language processing is performed on the preprocessed training data to obtain training data.

[0035] Exemplarily, the deduplication processing of the training data to be processed may include deduplication and denoising, wherein deduplication refers to removing data samples containing repeated information by comparing the information in the training data to be processed; denoising refers to removing abnormal data, noise data, etc. in the training data to be processed. The pre-processed training data obtained after deduplication processing can be further processed using natural language processing technology to obtain training data.

[0036] Natural Language Processing (NLP) can include lexical analysis, syntactic analysis, semantic understanding, feature extraction and other processing. Among them, lexical analysis can include word segmentation, part-of-speech tagging, named entity recognition, stop word removal and other operations. Word segmentation can divide continuous natural language text into individual words according to certain rules. Part-of-speech tagging can determine the part of speech of each word, such as noun, verb, adjective, etc. Named entity recognition can identify entities with specific meanings in the text, such as names of people, places, and organizational names. Syntactic analysis can include dependency analysis, phrase structure analysis, etc. Dependency analysis can analyze the dependency relationship between words in a sentence, determine which word is the core word and the modification, supplement and other relationships between other words and the core word. Phrase structure analysis can parse a sentence into different phrase structures and further analyze the grammatical structure of the sentence. Semantic understanding can include word sense disambiguation, semantic role labeling, text implication recognition, etc. Word sense disambiguation can determine the specific meaning of each word according to the context to avoid ambiguity caused by words with multiple meanings. Semantic role labeling can label the semantic roles played by each component in a sentence. Text entailment recognition can determine whether there is an entailment relationship between two texts.

[0037] In this embodiment, the training data to be processed is deduplicated to obtain preprocessed training data; natural language processing is performed on the preprocessed training data to obtain training data, which can improve the quality of the training data, thereby improving the training efficiency and improving the accuracy of the target question-answering model after training.

[0038] In some optional embodiments, the data processing method of this embodiment also includes: obtaining fiscal and taxation policy data of the current period according to a preset period; if there is a difference between the fiscal and taxation policy data of the current period and the fiscal and taxation policy data of the previous period, constructing an incremental data sample based on the fiscal and taxation policy data of the current period; based on the incremental data sample, performing online update training on the target question and answer model to obtain an updated target question and answer model.

[0039] Exemplarily, the preset period can be flexibly set by those skilled in the art according to actual conditions, and this embodiment does not limit this. According to the preset period, the fiscal and taxation policy data of the current period is obtained, that is, the latest fiscal and taxation policy data is obtained, for example, the latest fiscal and taxation policy data is regularly obtained from channels such as government official websites. Compare whether the fiscal and taxation policy data obtained in the current period is the same as the fiscal and taxation policy data obtained in the previous period. If there is a difference between the fiscal and taxation policy data of the current period and the fiscal and taxation policy data of the previous period, it means that the fiscal and taxation policy data is updated, then an incremental data sample is constructed based on the fiscal and taxation policy data of the current period, and then based on the incremental data sample, the target question and answer model is updated and trained online to obtain an updated target question and answer model.

[0040] Since the fiscal and taxation policy data will be continuously updated, this embodiment obtains the fiscal and taxation policy data of the current period by setting a preset period; if there is a difference between the fiscal and taxation policy data of the current period and the fiscal and taxation policy data of the previous period, an incremental data sample is constructed based on the fiscal and taxation policy data of the current period; based on the incremental data sample, the target question and answer model is updated and trained online, so that the target question and answer model can adapt to the update iteration of the fiscal and taxation policy data in a timely manner, thereby improving the accuracy of the target question and answer model.

[0041] In some optional embodiments, the data processing method of this embodiment also includes: obtaining test data, the test data including sample data pairs consisting of historical fiscal and taxation problem data samples and corresponding historical fiscal and taxation response data samples; based on the target question and answer model, inferring the historical fiscal and taxation problem data samples in the test data to obtain test fiscal and taxation response data; based on the test fiscal and taxation response data and the corresponding historical fiscal and taxation response data samples, determining an evaluation result for the target question and answer model; and optimizing the target question and answer model based on the evaluation result.

[0042] Exemplarily, a large number of sample data pairs consisting of historical fiscal and taxation problem data samples and corresponding historical fiscal and taxation response data samples can be obtained and divided into two parts, one part is training data and the other part is test data, that is, the sample data pairs consisting of historical fiscal and taxation problem data samples and corresponding historical fiscal and taxation response data samples included in the test data are different from the training data. Based on the target question-answering model, the historical fiscal and taxation problem data samples in the test data are inferred to obtain test fiscal and taxation response data, and then based on the test fiscal and taxation response data and the corresponding historical fiscal and taxation response data samples, the evaluation results for the target question-answering model are determined. The evaluation results may include evaluation indicators such as accuracy, precision, recall, and F1 value. Then, the target question-answering model can be optimized according to the evaluation results.

[0043] This embodiment can further optimize the target question and answer model through test data to improve the accuracy of the target question and answer results.

[0044] S108. Provide feedback on the financial and tax response data.

[0045] For example, refer to Figure 2 The interactive interface may include a result feedback area, and the financial and tax response data may be displayed in the result feedback area to provide feedback to the user, or the feedback may be provided in the form of voice broadcast, which is not limited in this embodiment. The interactive interface may also include other operation areas, which may include creating new dialogues, historical problem records, etc., which are not limited in this embodiment.

[0046] The embodiment of the present application obtains feature data in the financial and taxation question data by extracting features from the financial and taxation question data asked by the user; then the feature data is processed using a pre-trained target question-answering model to output financial and taxation response data without the need for manual answers, thereby improving the efficiency of answering financial and taxation questions raised by users and reducing labor costs.

[0047] In some optional embodiments, the data processing method of this embodiment also includes: in response to the user's second input operation, obtaining the user's evaluation information on the financial and taxation response data; analyzing the evaluation information and the financial and taxation question data and financial and taxation response data corresponding to the evaluation information to obtain analysis results; and optimizing the target question and answer model based on the analysis results.

[0048] For example, refer to Figure 2 Users can provide feedback in the result feedback area of ​​the provided interactive interface to evaluate the financial and tax response data output by the target question-and-answer model. The second input operation can be a click operation or a text input operation, such as clicking an evaluation button set in the interactive interface, filling in an evaluation message box set in the interactive interface, or filling in an evaluation in a questionnaire. This embodiment does not limit this.

[0049] The evaluation information may include an evaluation of the accuracy of the fiscal and taxation response data, which may be in the form of scoring (e.g., 1-5 points), selecting an evaluation level (e.g., excellent, good, average, poor, very poor), or providing a specific text description. The evaluation information and the fiscal and taxation problem data and fiscal and taxation response data corresponding to the evaluation information are recorded, and the evaluation information and the fiscal and taxation problem data and fiscal and taxation response data corresponding to the evaluation information are analyzed regularly. For example, if many users give a low accuracy evaluation under a certain type of specific fiscal and taxation problem data, it is necessary to focus on the model reasoning corresponding to this type of fiscal and taxation problem data. Through analysis, it is possible to analyze which fiscal and taxation problem data leads to the reasoning results with low accuracy, and in which aspects the gap between these reasoning results and user expectations is specifically reflected, that is, to obtain the analysis results. Based on the analysis results, the target question-answering model is optimized. For example, for those fiscal and taxation problem data that are pointed out by users to have low reasoning results, more data samples related to the fiscal and taxation problem data can be collected for supplementary training; or the architecture and parameters of the target question-answering model can be adjusted, such as increasing the number of layers and neurons of the network model.

[0050] In this embodiment, the evaluation information and the financial and taxation question data and financial and taxation response data corresponding to the evaluation information are analyzed to obtain analysis results; based on the analysis results, the target question and answer model is optimized to further improve the accuracy of the target question and answer model, thereby improving the user experience.

[0051] Reference Figure 3 , shows a structural block diagram of a data processing device according to an exemplary embodiment of the present application.

[0052] The data processing device of this embodiment includes an acquisition module 302 , a preprocessing module 304 , a model reasoning module 306 and an interaction module 308 .

[0053] Among them, the acquisition module 302 is used to respond to the user's first input operation and obtain the fiscal and taxation question data asked by the user; the preprocessing module 304 is used to extract features from the fiscal and taxation question data to obtain feature data in the fiscal and taxation question data; the model reasoning module 306 is used to process the feature data based on the pre-trained target question and answer model and output fiscal and taxation response data. The target question and answer model is a neural network model trained by sample data pairs consisting of historical fiscal and taxation question data samples and corresponding historical fiscal and taxation response data samples, as well as fiscal and taxation policy data samples; the interaction module 308 is used to feed back the fiscal and taxation response data.

[0054] In some optional embodiments, the data processing device of this embodiment also includes a model training module, which is used to: obtain training data to be processed; the training data includes sample data pairs consisting of historical fiscal and taxation problem data samples and corresponding historical fiscal and taxation response data samples, as well as fiscal and taxation policy data samples; perform feature extraction on the training data to be processed to obtain training data; based on the training data, perform supervised training on the preset question and answer model to be trained to obtain a target question and answer model.

[0055] In some optional embodiments, the model training module is also used to: obtain the fiscal and taxation policy data of the current period according to a preset period; if there is a difference between the fiscal and taxation policy data of the current period and the fiscal and taxation policy data of the previous period, construct an incremental data sample based on the fiscal and taxation policy data of the current period; based on the incremental data sample, perform online update training on the target question and answer model to obtain an updated target question and answer model.

[0056] In some optional embodiments, the preprocessing module 304 is further used to: perform deduplication processing on the training data to be processed to obtain preprocessed training data; and perform natural language processing on the preprocessed training data to obtain training data.

[0057] In some optional embodiments, the model training module is also used to: obtain test data, the test data including sample data pairs consisting of historical fiscal and taxation question data samples and corresponding historical fiscal and taxation response data samples; based on the target question and answer model, infer the historical fiscal and taxation question data samples in the test data to obtain test fiscal and taxation response data; based on the test fiscal and taxation response data and the corresponding historical fiscal and taxation response data samples, determine the evaluation results for the target question and answer model; and optimize the target question and answer model based on the evaluation results.

[0058] In some optional embodiments, the model training module is also used to: obtain the user's evaluation information on the financial and tax response data in response to the user's second input operation; analyze the evaluation information and the financial and tax question data and financial and tax response data corresponding to the evaluation information to obtain analysis results; and optimize the target question and answer model based on the analysis results.

[0059] The data processing device of this embodiment is used to implement the corresponding data processing methods in the aforementioned multiple method embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be described in detail here. In addition, the functional implementation of each module in the data processing device of this embodiment can refer to the description of the corresponding parts in the aforementioned method embodiments, which will not be described in detail here.

[0060] Reference Figure 4 , shows a schematic diagram of the structure of an electronic device according to an exemplary embodiment of the present application. The specific embodiment of the present invention does not limit the specific implementation of the electronic device.

[0061] like Figure 4 As shown, the electronic device may include: a processor (processor) 401 , a memory (memory) 403 , a communication bus 404 , and a communication interface (Communications Interface) 405 .

[0062] Wherein: the processor 401 , the memory 403 and the communication interface 405 communicate with each other via the communication bus 404 .

[0063] The communication interface 405 is used to communicate with other electronic devices or servers.

[0064] The processor 401 is used to execute the program 402, and specifically can execute the steps of any data processing method in the above embodiments.

[0065] Specifically, the program 402 may include program codes, which include computer operation instructions.

[0066] The processor 401 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application. The one or more processors included in the smart device may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0067] The memory 403 is used to store the program 402. The memory 403 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0068] The program 402 can be specifically used to enable the processor 401 to execute the steps of any data processing method described in the embodiment. The specific implementation of each step in the program 402 can refer to the corresponding description of the steps and units executed by any data processing method in the above steps, which will not be repeated here. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the above-described devices and modules can refer to the corresponding process description in the aforementioned method embodiment.

[0069] An embodiment of the present application further provides a computer storage medium on which a computer program is stored. When the program is executed by a processor, the data processing method described in any one of the above-mentioned method embodiments is implemented.

[0070] An embodiment of the present application also provides a computer program product, including computer instructions, which instruct a computing device to execute operations corresponding to the data processing method described in any one of the above-mentioned multiple method embodiments.

[0071] It should be pointed out that, according to the needs of implementation, the various components / steps described in the embodiments of the present application can be split into more components / steps, or two or more components / steps or partial operations of components / steps can be combined into new components / steps to achieve the purpose of the embodiments of the present application.

[0072] The above-mentioned method according to the embodiment of the present application can be implemented in hardware, firmware, or implemented as software or computer code that can be stored in a recording medium (such as a CD ROM, RAM, floppy disk, hard disk or magneto-optical disk), or implemented as a computer code originally stored in a remote recording medium or a non-temporary machine-readable medium downloaded through a network and to be stored in a local recording medium, so that the method described herein can be stored in such software processing on a recording medium using a general-purpose computer, a special-purpose processor or programmable or special-purpose hardware (such as an ASIC or FPGA). It can be understood that a computer, a processor, a microprocessor controller or programmable hardware includes a storage component (e.g., RAM, ROM, flash memory, etc.) that can store or receive software or computer code, and when the software or computer code is accessed and executed by a computer, a processor or hardware, the method described herein is implemented. In addition, when a general-purpose computer accesses the code for implementing the method shown here, the execution of the code converts the general-purpose computer into a special-purpose computer for executing the method shown here.

[0073] Those of ordinary skill in the art will appreciate that the units and method steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the embodiments of the present application.

[0074] The above implementation methods are only used to illustrate the embodiments of the present application, and are not limitations on the embodiments of the present application. Ordinary technicians in the relevant technical field can make various changes and modifications without departing from the spirit and scope of the embodiments of the present application. Therefore, all equivalent technical solutions also belong to the scope of the embodiments of the present application. The scope of patent protection of the embodiments of the present application should be limited by the claims.

Claims

1. A data processing method, characterized in that: include: In response to a first input operation of the user, obtaining financial and taxation question data asked by the user; Extracting features from the financial and taxation problem data to obtain feature data from the financial and taxation problem data; Processing the feature data based on a pre-trained target question-answering model to output fiscal and tax response data, wherein the target question-answering model is a neural network model trained through sample data pairs consisting of historical fiscal and tax question data samples and corresponding historical fiscal and tax response data samples, as well as fiscal and tax policy data samples; The financial and tax response data will be fed back.

2. The data processing method according to claim 1, characterized in that: The training process of the target question answering model includes: Get the training data to be processed; Extracting features from the training data to be processed to obtain training data; the training data includes sample data pairs consisting of historical fiscal and taxation problem data samples and corresponding historical fiscal and taxation response data samples, as well as fiscal and taxation policy data samples; Based on the training data, supervised training is performed on the preset question-answering model to be trained to obtain a target question-answering model.

3. The data processing method according to claim 2, characterized in that: The method further comprises: Obtain the fiscal and tax policy data for the current period according to the preset period; If the fiscal and taxation policy data of the current period differs from the fiscal and taxation policy data of the previous period, constructing an incremental data sample based on the fiscal and taxation policy data of the current period; Based on the incremental data samples, the target question-answering model is updated and trained online to obtain an updated target question-answering model.

4. The data processing method according to claim 2 or 3, characterized in that: The step of extracting features from the training data to be processed to obtain the training data includes: Performing deduplication processing on the training data to be processed to obtain preprocessed training data; Natural language processing is performed on the preprocessed training data to obtain training data.

5. The data processing method according to claim 2, characterized in that: The method further comprises: Acquire test data, wherein the test data includes a sample data pair consisting of a historical fiscal and taxation question data sample and a corresponding historical fiscal and taxation answer data sample; Based on the target question-answering model, reasoning is performed on the historical finance and taxation question data samples in the test data to obtain test finance and taxation response data; Determine an evaluation result for the target question-answering model based on the test fiscal and taxation response data and the corresponding historical fiscal and taxation response data samples; The target question-answering model is optimized based on the evaluation result.

6. The data processing method according to claim 1, characterized in that: The method further comprises: In response to a second input operation of the user, obtaining evaluation information of the user on the financial and tax reply data; Analyze the evaluation information and the financial and taxation problem data and the financial and taxation response data corresponding to the evaluation information to obtain an analysis result; Based on the analysis results, the target question-answering model is optimized.

7. A data processing device, characterized in that: include: An acquisition module, for acquiring the financial and taxation question data asked by the user in response to a first input operation of the user; A preprocessing module, used for extracting features from the financial and taxation problem data to obtain feature data from the financial and taxation problem data; A model reasoning module, used to process the feature data based on a pre-trained target question-answering model and output fiscal and taxation response data, wherein the target question-answering model is a neural network model trained through sample data pairs consisting of historical fiscal and taxation question data samples and corresponding historical fiscal and taxation response data samples, as well as fiscal and taxation policy data samples; The interactive module is used to feed back the financial and tax response data.

8. An electronic device, characterized in that: include: A processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform an operation corresponding to the method according to any one of claims 1 to 6.

9. A computer storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

10. A computer program product, characterized in that The method comprises computer instructions, wherein the computer instructions instruct a computing device to execute operations corresponding to the method according to any one of claims 1 to 6.

Images