CN113269433A - Tax risk prediction method, apparatus, medium, and computer program product - Google Patents
Tax risk prediction method, apparatus, medium, and computer program product Download PDFInfo
- Publication number
- CN113269433A CN113269433A CN202110553379.9A CN202110553379A CN113269433A CN 113269433 A CN113269433 A CN 113269433A CN 202110553379 A CN202110553379 A CN 202110553379A CN 113269433 A CN113269433 A CN 113269433A
- Authority
- CN
- China
- Prior art keywords
- tax
- risk prediction
- target
- training
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 60
- 238000004590 computer program Methods 0.000 title claims description 14
- 238000000605 extraction Methods 0.000 claims abstract description 80
- 238000013058 risk prediction model Methods 0.000 claims abstract description 74
- 238000012549 training Methods 0.000 claims description 149
- 238000011156 evaluation Methods 0.000 claims description 48
- 230000002159 abnormal effect Effects 0.000 claims description 14
- 230000000052 comparative effect Effects 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims 1
- 238000001514 detection method Methods 0.000 claims 1
- 238000005457 optimization Methods 0.000 claims 1
- 238000012502 risk assessment Methods 0.000 abstract description 13
- 238000004891 communication Methods 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 4
- 230000005856 abnormality Effects 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 3
- 239000000654 additive Substances 0.000 description 2
- 230000000996 additive effect Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000011478 gradient descent method Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/10—Tax strategies
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/12—Accounting
- G06Q40/123—Tax preparation or submission
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Theoretical Computer Science (AREA)
- Finance (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Technology Law (AREA)
- Software Systems (AREA)
- Game Theory and Decision Science (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Educational Administration (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application discloses a tax risk prediction method, which comprises the following steps: acquiring a tax associated data set corresponding to a target user, and performing feature extraction on the tax associated data set based on a target feature extraction model to obtain a target tax feature set, wherein the feature extraction model is obtained by performing comparison learning based on preset tampered tax associated data and preset non-tampered tax associated data; and performing tax risk prediction on the target user based on a tax risk prediction model and the target tax feature set to obtain a tax risk prediction result. The application solves the technical problem that the tax risk analysis accuracy is low.
Description
Technical Field
The present application relates to the field of artificial intelligence technologies, and in particular, to a tax risk prediction method, apparatus, medium, and computer program product.
Background
With the continuous development of computer software, artificial intelligence and big data cloud service application, the application of artificial intelligence technology is more and more extensive. Tax status is an important information for enterprise development. At present, the tax associated data of an enterprise is generally analyzed manually by financial staff to judge the tax risk of the enterprise, but because the tax risk analysis is often associated with the subjectivity of the financial staff, the tax risk analysis result has a larger error, and the accuracy of the tax risk analysis is lower.
Disclosure of Invention
The present application mainly aims to provide a tax risk prediction method, device, medium, and computer program product, which aim to solve the technical problem of low accuracy of tax risk analysis in the prior art.
In order to achieve the above object, the present application provides a tax risk prediction method, which is applied to a tax risk prediction device, and the tax risk prediction method includes:
acquiring a tax associated data set corresponding to a target user, and performing feature extraction on the tax associated data set based on a target feature extraction model to obtain a target tax feature set, wherein the feature extraction model is obtained by performing comparison learning based on preset tampered tax associated data and preset non-tampered tax associated data;
and performing tax risk prediction on the target user based on a tax risk prediction model and the target tax feature set to obtain a tax risk prediction result.
The application also provides a tax risk prediction device, tax risk prediction device is virtual device, just tax risk prediction device is applied to tax risk prediction equipment, tax risk prediction device includes:
the system comprises a characteristic extraction module, a characteristic extraction module and a characteristic extraction module, wherein the characteristic extraction module is used for acquiring a tax-related data set corresponding to a target user, and extracting characteristics of the tax-related data set based on a target characteristic extraction model to acquire a target tax characteristic set, and the characteristic extraction model is obtained by performing comparison learning based on preset tampered tax-related data and preset untampered tax-related data;
and the tax risk prediction module is used for predicting the tax risk of the target user based on a tax risk prediction model and the target tax feature set to obtain a tax risk prediction result.
The application also provides a tax risk prediction device, tax risk prediction device is entity device, tax risk prediction device includes: a memory, a processor, and a program of the tax risk prediction method stored on the memory and executable on the processor, the program of the tax risk prediction method when executed by the processor may implement the steps of the tax risk prediction method as described above.
The present application also provides a readable storage medium having stored thereon a program for implementing a tax risk prediction method, the program implementing the steps of the tax risk prediction method as described above when executed by a processor.
The present application also provides a computer program product comprising a computer program which, when executed by a processor, performs the steps of the tax risk prediction method as described above.
The application provides a tax risk prediction method, equipment, medium and computer program product, compared with the technical means of manually analyzing tax associated data of an enterprise by financial staff to judge the tax risk of the enterprise in the prior art, the application firstly obtains a tax associated data set corresponding to a target user, and performs characteristic extraction on the tax associated data set based on a target characteristic extraction model to obtain a target tax characteristic set, wherein, as the characteristic extraction model is obtained by performing comparison learning based on preset tampered tax associated data and preset non-tampered tax associated data, the distance between tax characteristics corresponding to intra-class data in the preset tampered tax associated data and the preset non-tampered tax associated data can be drawn closer, and the distance between tax characteristics corresponding to inter-class data is drawn farther, so that the tax characteristics in the target tax characteristic set have class information, and then based on a tax risk prediction model and the target tax feature set, carrying out tax risk prediction on the target user to obtain a tax risk prediction result, namely, the target tax feature set with category information can be realized, and the tax risk prediction of the target user is carried out, so that more decision bases are provided for the tax risk prediction, and the automatic and accurate tax risk prediction is realized, thereby overcoming the technical defect that the accuracy of tax risk analysis is lower due to the fact that tax risk analysis is often associated with the subjectivity of financial staff, the tax risk analysis result has larger error, and the accuracy of tax risk analysis is lower, and improving the accuracy of tax risk prediction.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and together with the description, serve to explain the principles of the application.
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly described below, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
FIG. 1 is a schematic flow chart illustrating a tax risk prediction method according to a first embodiment of the present application;
FIG. 2 is a schematic flow chart illustrating a tax risk prediction method according to a second embodiment of the present application;
fig. 3 is a schematic device structure diagram of a hardware operating environment related to a tax risk prediction method in an embodiment of the present application.
The objectives, features, and advantages of the present application will be further described with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
In a first embodiment of the tax risk prediction method, referring to fig. 1, the tax risk prediction method includes:
step S10, acquiring a tax associated data set corresponding to a target user, and performing feature extraction on the tax associated data set based on a target feature extraction model to acquire a target tax feature set, wherein the feature extraction model is obtained by performing comparative learning based on preset tampered tax associated data and preset untampered tax associated data;
in this embodiment, it should be noted that the target user may be an enterprise user or a personal user, the tax-related data set at least includes tax-related data of a preset data type, and the target tax feature set at least includes a target tax characteristic corresponding to the tax-related data. The preset data type may be set to determine a data type according to a type of the data itself, for example, bank flow type data, invoice type data, payment instrument flow type data, and the like, that is, may be data determined according to a data source corresponding to the data, for example, the preset data type a is a type to which data from the data source a (bank) belongs, and the preset data type B is a type to which data from the data source B (payment instrument) belongs.
The method comprises the steps of obtaining a tax associated data set corresponding to a target user, carrying out feature extraction on the tax associated data set based on a target feature extraction model, and obtaining a target tax feature set, wherein the feature extraction model is obtained by comparison and learning based on preset tampered tax associated data and preset untampered tax associated data, specifically, obtaining tax associated data corresponding to each preset data type corresponding to the target user, and further carrying out feature extraction on each tax associated data respectively based on a target feature extraction model corresponding to each preset data type, so as to obtain target tax features corresponding to each tax associated data.
Before the step of performing feature extraction on the tax related data set based on the target feature extraction model to obtain a target tax feature set, wherein the feature extraction model is obtained by performing comparative learning based on preset tampered tax related data and preset untampered tax related data, the tax risk prediction method further includes:
step A10, obtaining a training sample and a sample tampering identifier corresponding to the training sample, and extracting a positive tax data sample and a negative tax data sample corresponding to the training sample from the preset tampering tax associated data and the preset non-tampering tax associated data based on the sample tampering identifier;
in this embodiment, it should be noted that the sample tampering flag is a flag indicating whether the training sample is tampered, for example, if the sample tampering flag is set to 1, it indicates that the training sample is tampered, and if the sample tampering flag is set to 0, it indicates that the training sample is not tampered.
Obtaining a training sample and a sample tampering identifier corresponding to the training sample, extracting a positive tax data sample and a negative tax data sample corresponding to the training sample from the preset tampering tax-related data and the preset non-tampering tax-related data based on the sample tampering identifier, specifically, obtaining a training sample and a sample tampering identifier corresponding to the training sample, and judging whether the training sample is tampered or not based on the sample tampering identifier, if the training sample is tampered, randomly selecting a positive sample from the preset tampering tax-related data as the positive tax data sample corresponding to the training sample, randomly selecting a preset number of negative samples from the preset non-tampering tax-related data as the negative tax data sample corresponding to the training sample, and if the training sample is not tampered, randomly selecting a preset number of negative examples of samples from the preset tampered tax related data as negative examples of tax data samples corresponding to the training samples, and randomly selecting positive examples of samples from the preset untampered tax related data as positive examples of tax data samples corresponding to the training samples.
A20, extracting the features of the training sample based on a feature extraction model to be trained to obtain training tax features corresponding to the training sample;
in this embodiment, it should be noted that the feature extraction model to be trained is an untrained target feature extraction model.
The training method comprises the steps of extracting features of a training sample based on a feature extraction model to be trained to obtain training tax features corresponding to the training sample, specifically, inputting the training sample into the feature extraction model to be trained to map the training sample to a preset feature dimension, extracting the features of the training sample, and obtaining the training tax features corresponding to the training sample.
Step A30, respectively performing feature extraction on the positive tax data sample and the negative tax data sample based on a preset prior feature extraction model to obtain a positive tax feature corresponding to the positive tax data sample and a negative tax feature corresponding to the negative tax data sample;
in this embodiment, it should be noted that the preset prior feature extraction model is a pre-trained feature extraction model, and the number of negative tax data samples is at least 1.
Respectively extracting the characteristics of the positive tax data sample and the negative tax data sample based on a preset prior characteristic extraction model to obtain positive tax characteristics corresponding to the positive tax data sample and negative tax characteristics corresponding to the negative tax data sample, specifically, the positive tax data samples are input into a preset prior characteristic extraction model to map the positive tax data samples to preset characteristic dimensions, performing feature extraction on the positive tax data sample to obtain positive tax features corresponding to the positive tax data sample, and in the same way, negative case tax data samples are input into a preset prior characteristic extraction model to map the negative case tax data samples to preset characteristic dimensions, and performing feature extraction on the negative case tax data sample to obtain a negative case tax feature corresponding to the negative case tax data sample.
Step A40, constructing a contrast learning loss based on the difference between the training sample characteristics and the positive example tax characteristics and the difference between the training sample characteristics and the negative example tax characteristics;
in this embodiment, based on the difference between the training sample feature and the positive example tax feature and the difference between the training sample feature and the negative example tax feature, a comparison learning loss is constructed through a preset comparison learning loss calculation formula, where the preset comparison learning loss calculation formula is specifically as follows:
wherein L is the contrast learning loss, uAFor the training sample features, uBFor the purposes of the positive example tax feature,
for the negative tax characteristics, M is the number of the negative tax characteristics, and further when the distance between the positive tax characteristics and the training sample characteristics is small enough and the distance between each negative tax characteristics and the training sample characteristics is large enough, the comparison learning loss can be converged, and further the target characteristic extraction model updated based on the comparison learning loss can have the capability of zooming in the distance between the tax characteristics and the positive tax characteristics as the positive examples and zooming out the distance between the tax characteristics and the negative tax characteristics as the negative examples, and further the target characteristic extraction model can generate different tax characteristics based on samples of different sample types (positive examples or negative examples), so that the generated tax characteristics have sample category information, the information content contained in the tax characteristics generated by characteristic extraction is improved, and further more decision bases can be provided for tax risk prediction, the accuracy of tax risk prediction is improved, and meanwhile, the knowledge of the preset prior feature extraction model can be promoted to be learned by the feature extraction model to be trained, so that the feature extraction model to be trained can be converged more quickly in the training process, and the training efficiency of the feature extraction model is improved.
And A50, optimizing the feature extraction model to be trained based on the comparison learning loss to obtain the target feature extraction model.
In this embodiment, the feature extraction model to be trained is optimized based on the comparison learning loss to obtain the target feature extraction model, specifically, whether the comparison learning loss is converged is determined, if the comparison learning loss is converged, the feature extraction model to be trained is used as the target feature extraction model, and if the comparison learning loss is not converged, the feature extraction model to be trained is updated based on a model gradient calculated based on the comparison learning loss by a preset model updating method, and the step of obtaining the training sample and the sample tampering identifier corresponding to the training sample is returned to be executed, where the preset model updating method includes a gradient ascent method and a gradient descent method.
And step S20, based on the tax risk prediction model and the target tax characteristic set, carrying out tax risk prediction on the target user to obtain a tax risk prediction result.
In this embodiment, it should be noted that the target tax characteristic set at least includes a target tax characteristic.
Performing tax risk prediction on the target user based on a tax risk prediction model and the target tax feature set to obtain a tax risk prediction result, specifically, performing weighted combination on each target tax feature based on each preset initial combination weight to obtain a combined tax feature, performing model prediction by inputting the combined tax feature into the tax risk prediction model, performing two classifications on the combined tax feature to obtain two classification results, further, determining whether the target user has a high tax risk based on the two classification results to obtain a tax risk prediction result, for example, setting the two classification results as a classification A or a classification B, further, if the two classification results are a classification A, determining that the target user has a high tax risk, and if the tax risk prediction result is a high tax risk, and if the two classification results are a classification B, and judging that the target user has a low tax risk, wherein the tax risk prediction result is the low tax risk.
The embodiment of the application provides a tax risk prediction method, compared with the technical means that a financial worker manually analyzes tax-related data of an enterprise to judge the tax risk of the enterprise in the prior art, the tax risk prediction method firstly acquires a tax-related data set corresponding to a target user and performs characteristic extraction on the tax-related data set based on a target characteristic extraction model to acquire a target tax characteristic set, wherein the characteristic extraction model is obtained by performing comparison learning based on preset tampered tax-related data and preset untampered tax-related data, the distance between tax characteristics corresponding to intra-class data in the preset tampered tax-related data and the preset untampered tax-related data can be shortened, and the distance between tax characteristics corresponding to inter-class data is lengthened, so that the tax characteristics in the target tax characteristic set have class information, and then based on a tax risk prediction model and the target tax feature set, carrying out tax risk prediction on the target user to obtain a tax risk prediction result, namely, the target tax feature set with category information can be realized, and the tax risk prediction of the target user is carried out, so that more decision bases are provided for the tax risk prediction, and the automatic and accurate tax risk prediction is realized, thereby overcoming the technical defect that the accuracy of tax risk analysis is lower due to the fact that tax risk analysis is often associated with the subjectivity of financial staff, the tax risk analysis result has larger error, and the accuracy of tax risk analysis is lower, and improving the accuracy of tax risk prediction.
Further, referring to fig. 2, based on the first embodiment of the present application, in another embodiment of the present application, the step of predicting the tax risk of the target user based on the tax risk prediction model and the target tax feature set to obtain a tax risk prediction result includes:
step S21, based on the tax risk prediction model and the target tax feature set, carrying out tax risk prediction on the target user to obtain an initial tax risk prediction result;
in this embodiment, based on the tax risk prediction model and the target tax characteristic set, tax risk prediction is performed on the target user to obtain an initial tax risk prediction result, specifically, based on each preset initial combination weight, weighted combination is performed on each target tax characteristic in the target tax characteristic set to obtain a combined tax characteristic, and then the combined tax characteristic is input into the tax risk prediction model to perform secondary classification on the combined tax characteristic to obtain a secondary classification result, and then based on the secondary classification result, whether the target user has a high tax risk is determined to obtain a tax risk prediction result.
Step S22, performing model interpretation aiming at the tax risk prediction model on the initial tax risk prediction result to evaluate the tampering degree of each tax-related data and obtain a tampering degree evaluation result;
in this embodiment, the model interpretation specific to the tax risk prediction model is performed on the initial tax risk prediction result to evaluate the tampering degree of each tax-related data, so as to obtain a tampering degree evaluation result, specifically, the model interpretation specific to the tax risk prediction model is performed on the initial tax risk prediction result to calculate the feature contribution degree of each target tax feature to the initial tax risk prediction result generated by the tax risk prediction model, and then the tampering degree of the tax-related data corresponding to each target tax feature is evaluated based on each feature contribution degree, so as to obtain a tampering degree evaluation result.
Wherein the tax-related data set at least comprises tax-related data of a preset data type, the target tax characteristic set at least comprises a target tax characteristic corresponding to the tax-related data,
the model interpretation of the initial tax risk prediction result for the tax risk prediction model is performed to evaluate the tampering degree of each tax-related data, and the step of obtaining the tampering degree evaluation result includes:
step S221, respectively calculating the feature contribution degree of each target tax feature to the initial tax risk prediction result based on the tax risk prediction model;
in this embodiment, it should be noted that the feature contribution degree is a degree of influence of the target tax feature on the initial tax risk prediction result, where the feature contribution degree includes a positive feature contribution degree and a negative feature contribution degree, where the positive feature contribution degree indicates that the initial tax risk prediction result has a positive influence, the negative feature contribution degree indicates that the initial tax risk prediction result has a negative influence, for example, if the initial tax risk prediction result is a high tax risk, the positive feature contribution degree indicates that a positive incentive effect is provided for the promotion of the tax risk, and the negative feature contribution degree indicates that a negative incentive effect is provided for the promotion of the tax risk.
Respectively calculating the feature contribution of each target tax characteristic to the initial tax risk prediction result based on the tax risk prediction Model, specifically, respectively calculating the feature contribution of each target tax characteristic to the initial tax risk prediction result in a preset feature contribution calculation mode based on the tax risk prediction Model, where the preset feature contribution calculation mode includes SHAP (SHAPLE Additive Explanation, ShaPLE Additive Model interpretation) and LIME (Local Interpredictive Model-aggregate Explanation, Model-independent Local interpretation), and the like.
Step S222, based on each feature contribution degree, evaluating a tampering degree of each tax related data, and obtaining a tampering degree evaluation result.
In this embodiment, based on each of the feature contribution degrees, a tampering degree of each of the tax related data is evaluated to obtain the tampering degree evaluation result, and specifically, based on a value distribution of each of the feature contribution degrees, whether an abnormal feature contribution degree exists in each of the feature contribution degrees is determined, if the abnormal feature contribution degree exists, it is determined that the tax related data corresponding to the abnormal feature contribution degree is tampered, the tax related data corresponding to the feature contribution degrees other than the abnormal feature contribution degree is not tampered, that is, the tampering degree is 0, and further based on a value size of the abnormal feature contribution degree, the tampering degree of the corresponding tax related data is evaluated to obtain a tampering degree evaluation result corresponding to each of the tax related data, where the tampering degree evaluation result may be represented by a tampering degree evaluation vector, and a value on each bit in the tampering degree evaluation vector is a value corresponding to one tax related data representing a tampering degree corresponding to the tampering degree For example, if the tamper evaluation vector is (0, 0.9), 0 indicates that the degree of tampering of the tax related data type a is 0, and 0.9 indicates that the degree of tampering of the tax related data type B is 90%.
Wherein, the step of evaluating the tampering degree of each tax related data based on each feature contribution degree, and the step of obtaining the result of evaluating the tampering degree comprises:
step B10, based on a preset abnormity discrimination model, carrying out abnormity characteristic contribution degree discrimination on each characteristic contribution degree to obtain an abnormity discrimination result;
in this embodiment, it should be noted that the preset anomaly determination model is a preset machine learning model for determining whether value distribution anomaly exists in each feature contribution degree.
Performing abnormal feature contribution degree discrimination on each feature contribution degree based on a preset abnormal discrimination model to obtain an abnormal discrimination result, specifically, splicing each feature contribution degree into a feature contribution degree vector consisting of each feature contribution degree, and executing model discrimination by inputting the feature contribution vector into a preset anomaly discrimination model, the method comprises the steps of judging the contribution degree of each feature to obtain the result of judging the abnormality, and in an implementable mode, the abnormal judgment result may be set as a vector consisting of 0 and other values different from 0, 0 indicates that the feature contribution degree corresponding to the bit is not the abnormal feature contribution degree, the other numerical values not equal to 0 indicate that the characteristic contribution degree corresponding to the bit is the abnormal characteristic contribution degree, and the size of the other numerical values not equal to 0 indicates the tampering degree of the corresponding financial data.
And step B20, based on the abnormal judgment result, evaluating the tampering degree of each tax related data to obtain a tampering degree evaluation result.
In this embodiment, based on the abnormality determination result, the tampering degree of each piece of financial data is evaluated to obtain a tampering degree evaluation result, and specifically, based on the value of each numerical value in the abnormality determination result, the tampering degree of each piece of financial data is determined to obtain a tampering degree evaluation result, where the tampering degree evaluation result includes the tampering degree of each piece of financial data.
And step S23, performing secondary tax risk prediction on the target user based on the tampering degree evaluation result, the tax risk prediction model and the target tax feature set to obtain a target tax risk prediction result.
In this embodiment, based on the tampering degree evaluation result, the tax risk prediction model and the target tax feature set, performing secondary tax risk prediction on the target user to obtain a target tax risk prediction result, specifically, based on the tampering degree evaluation result, updating each preset initial combination weight to obtain each target combination weight, and based on each target combination weight, performing weighted concatenation on each target tax feature to obtain a target combination tax feature, and then performing model prediction by inputting the target combination tax feature into the tax risk prediction model to perform model prediction, performing secondary tax risk prediction on the target user to obtain a target tax risk prediction result.
Wherein the target tax characteristic set at least comprises a target tax characteristic,
the step of performing secondary tax risk prediction on the target user based on the tampering degree evaluation result, the tax risk prediction model and the target tax feature set to obtain a target tax risk prediction result comprises:
step S231, based on the tampering degree evaluation result, generating a target combination weight corresponding to each target tax characteristic;
in this embodiment, a target combination weight corresponding to each target tax characteristic is generated based on the tampering degree evaluation result, specifically, a preset initial combination weight corresponding to each target tax characteristic is optimized based on a size of a tampering degree representation value corresponding to each target tax characteristic in the tampering degree evaluation result, so as to obtain each target combination weight, for example, if the tampering degree representation value is 0.5 and the preset initial combination weight is 0.6, the target combination weight is 0.6 (1-0.5) ═ 0.3.
Step S232, based on each target combination weight, performing weighted combination on each target tax characteristic to obtain a target combination tax characteristic;
in this embodiment, each target tax characteristic is weighted and combined based on each target combination weight to obtain a target combination tax characteristic, specifically, each target tax characteristic corresponding to each target combination weight is weighted and spliced based on each target combination weight to obtain a target combination tax characteristic, for example, assuming that each target combination weight is 0.1, 0.5 and 0.4, each corresponding target tax characteristic is A, B and C, respectively, and further, the target combination tax characteristic is 0.1A spliced to 0.5B, and then spliced to 0.4C.
And step S233, performing secondary tax risk prediction on the target user based on the target combined tax characteristics and the tax risk prediction model to obtain a target tax risk prediction result.
In this embodiment, a secondary tax risk prediction is performed on the target user based on the target combined tax characteristic and the tax risk prediction model to obtain the target tax risk prediction result, specifically, the target combined tax characteristic is input into the tax risk prediction model to perform secondary classification on the target combined tax characteristic to obtain a secondary classification result, and then based on the secondary classification result, whether the target user has a high tax risk is determined to obtain the target tax risk prediction result.
Before the step of predicting the tax risk of the target user based on the tax risk prediction model and the target tax feature set and obtaining a tax risk prediction result, the tax risk prediction method further includes:
step C10, acquiring training tax associated data of each preset data type corresponding to a training user and corresponding real labels, and generating training tax characteristics corresponding to each training tax associated data;
in this embodiment, it should be noted that the real label is a label for identifying the tax risk of the training user, and is used to identify whether the tax risk of the training user is high or low.
The method comprises the steps of obtaining training tax associated data and corresponding real labels of preset data types corresponding to training users, and generating training tax characteristics corresponding to the training tax associated data, specifically, obtaining the training tax associated data and corresponding real labels of the preset data types corresponding to the training users, and respectively extracting the characteristics of the corresponding training tax associated data based on a characteristic extraction model corresponding to each preset data type, so as to obtain the training tax characteristics corresponding to the training tax associated data.
Step C20, based on the tax risk prediction model to be trained and each training tax characteristic, carrying out tax risk prediction on the training user to obtain an initial training prediction result;
in this embodiment, based on the tax risk prediction model to be trained and each training tax characteristic, the tax risk prediction is performed on the training user to obtain an initial training prediction result, specifically, based on a preset initial combination weight, each training tax characteristic is combined into an initial training combination tax characteristic in a weighted manner, and then the initial training combination tax characteristic is input into the tax risk prediction model to be trained to perform model prediction, so that the tax risk prediction is performed on the training user to obtain an initial training prediction result.
Step C30, performing model interpretation aiming at the tax risk prediction model to be trained on the initial training prediction result to evaluate the tampering degree of the tax-related data to obtain a training tampering degree evaluation result;
in this embodiment, the model interpretation of the initial training prediction result for the tax risk prediction model to be trained is performed, so as to evaluate the tampering degree of each training tax associated data, and obtain a training tampering degree evaluation result, specifically, performing model interpretation aiming at the tax risk prediction model to be trained on the initial training prediction result, respectively calculating the characteristic contribution degree of each training tax characteristic to the tax risk prediction model to be trained to generate the initial training prediction result, further evaluating the tampering degree of the training tax associated data corresponding to each training tax characteristic based on the characteristic contribution degree corresponding to each training tax characteristic to obtain a training tampering degree evaluation result, the process of performing model interpretation may specifically refer to the content in step S22 and its refinement step, which is not described herein again.
Step C40, performing secondary tax risk prediction on the training user based on the training tampering degree evaluation result, the tax risk prediction model to be trained and each training tax characteristic to obtain a target training prediction result;
in this embodiment, based on the training tampering degree evaluation result, the tax risk prediction model to be trained, and each training tax characteristic, performing secondary tax risk prediction on the training user to obtain a target training prediction result, specifically, based on the training tampering degree evaluation result, updating each preset initial combination weight to obtain each training combination weight, and based on each training combination weight, performing weighted concatenation on each training tax characteristic to obtain a training combination tax characteristic, and then performing model prediction by inputting the training combination financial characteristic into the tax risk prediction model to be trained, performing secondary financial condition prediction on the training user to obtain a target training prediction result.
Step C50, calculating the total model loss based on the real label, the initial training prediction result and the target training prediction result;
in this embodiment, a total model loss is calculated based on the real label, the initial training prediction result, and the target training prediction result, specifically, a first model loss is calculated based on a difference between the real label and the initial training prediction result, a second model loss is calculated based on a difference between the real label and the target training prediction result, and then the first model loss and the second model loss are weighted and summed to obtain the total model loss.
And step C60, optimizing the tax risk prediction model to be trained based on the model total loss to obtain the tax risk prediction model.
In this embodiment, the tax risk prediction model to be trained is optimized based on the total model loss to obtain the tax risk prediction model, specifically, whether the total model loss is converged is determined, if the total model loss is converged, the tax risk prediction model to be trained is used as the tax risk prediction model, and if the total model loss is not converged, the tax risk prediction model to be trained is updated based on a model gradient calculated based on the total model loss by a preset model updating method, and the step of obtaining training tax related data of each preset data type corresponding to a training user and a corresponding real label is returned, where the preset model optimizing method includes a gradient descent method, a gradient ascent method, and the like.
The embodiment of the application provides a method for predicting tax risk based on model interpretation, namely, firstly, based on the tax risk prediction model and the target tax characteristic set, carrying out tax risk prediction on a target user to obtain an initial tax risk prediction result, further carrying out model interpretation aiming at the tax risk prediction model on the initial tax risk prediction result to evaluate the tampering degree of each tax-related data and obtain a tampering degree evaluation result, so as to achieve the purpose of judging whether tax-related data is tampered and the tampering degree based on the model interpretation, further based on the tampering degree evaluation result, the tax risk prediction model and the target tax characteristic set, carrying out secondary tax risk prediction on the target user to obtain a target tax risk prediction result, and combining the tampering degree evaluation result, the tax risk of the target user is forecasted again, the situation that the tax risk forecasting accuracy becomes low due to tampering of tax related data is avoided, and the accuracy of tax risk forecasting is improved.
Referring to fig. 3, fig. 3 is a schematic device structure diagram of a hardware operating environment according to an embodiment of the present application.
As shown in fig. 3, the tax risk prediction device may include: a processor 1001, such as a CPU, a memory 1005, and a communication bus 1002. The communication bus 1002 is used for realizing connection communication between the processor 1001 and the memory 1005. The memory 1005 may be a high-speed RAM memory or a non-volatile memory (e.g., a magnetic disk memory). The memory 1005 may alternatively be a memory device separate from the processor 1001 described above.
Optionally, the tax risk prediction device may further include a rectangular user interface, a network interface, a camera, RF (Radio Frequency) circuitry, a sensor, audio circuitry, a WiFi module, and so on. The rectangular user interface may comprise a Display screen (Display), an input sub-module such as a Keyboard (Keyboard), and the optional rectangular user interface may also comprise a standard wired interface, a wireless interface. The network interface may optionally include a standard wired interface, a wireless interface (e.g., WI-FI interface).
Those skilled in the art will appreciate that the tax risk prediction device configuration shown in fig. 3 does not constitute a limitation of the tax risk prediction device and may include more or less components than those shown, or some components in combination, or a different arrangement of components.
As shown in fig. 3, a memory 1005, which is a kind of computer storage medium, may include an operating system, a network communication module, and a tax risk prediction program therein. The operating system is a program that manages and controls the hardware and software resources of the tax risk prediction device, supporting the operation of the tax risk prediction program as well as other software and/or programs. The network communication module is used for realizing communication among the components in the storage 1005 and communication with other hardware and software in the tax risk prediction system.
In the tax risk prediction apparatus shown in fig. 3, the processor 1001 is configured to execute a tax risk prediction program stored in the memory 1005 to implement the steps of the tax risk prediction method according to any one of the above-mentioned embodiments.
The specific implementation of the tax risk prediction device is basically the same as that of the embodiments of the tax risk prediction method, and is not described herein again.
The embodiment of the present application further provides a tax risk prediction device, tax risk prediction device is applied to tax risk prediction equipment, tax risk prediction device includes:
the system comprises a characteristic extraction module, a characteristic extraction module and a characteristic extraction module, wherein the characteristic extraction module is used for acquiring a tax-related data set corresponding to a target user, and extracting characteristics of the tax-related data set based on a target characteristic extraction model to acquire a target tax characteristic set, and the characteristic extraction model is obtained by performing comparison learning based on preset tampered tax-related data and preset untampered tax-related data;
and the tax risk prediction module is used for predicting the tax risk of the target user based on a tax risk prediction model and the target tax feature set to obtain a tax risk prediction result.
Optionally, the tax risk prediction module is further configured to:
based on the tax risk prediction model and the target tax characteristic set, carrying out tax risk prediction on the target user to obtain an initial tax risk prediction result;
performing model interpretation aiming at the tax risk prediction model on the initial tax risk prediction result to evaluate the tampering degree of each tax-related data to obtain a tampering degree evaluation result;
and performing secondary tax risk prediction on the target user based on the tampering degree evaluation result, the tax risk prediction model and the target tax feature set to obtain a target tax risk prediction result.
Optionally, the tax risk prediction module is further configured to:
generating target combination weights corresponding to the target tax characteristics based on the tampering degree evaluation result;
based on each target combination weight, performing weighted combination on each target tax characteristic to obtain a target combination tax characteristic;
and performing secondary tax risk prediction on the target user based on the target combined tax characteristics and the tax risk prediction model to obtain a target tax risk prediction result.
Optionally, the tax risk prediction module is further configured to:
respectively calculating the characteristic contribution degree of each target tax characteristic to the initial tax risk prediction result based on the tax risk prediction model;
and evaluating the tampering degree of the tax related data based on the characteristic contribution degrees to obtain the tampering degree evaluation result.
Optionally, the tax risk prediction module is further configured to:
based on a preset abnormity discrimination model, carrying out abnormity characteristic contribution degree discrimination on each characteristic contribution degree to obtain an abnormity discrimination result;
and evaluating the tampering degree of each tax-related data based on the abnormal judgment result to obtain a tampering degree evaluation result.
Optionally, the tax risk prediction device is further configured to:
acquiring a training sample and a sample tampering identifier corresponding to the training sample, and extracting a positive tax data sample and a negative tax data sample corresponding to the training sample from the preset tampering tax associated data and the preset non-tampering tax associated data based on the sample tampering identifier;
performing feature extraction on the training sample based on a feature extraction model to be trained to obtain training tax features corresponding to the training sample;
respectively extracting the characteristics of the positive tax data sample and the negative tax data sample based on a preset prior characteristic extraction model to obtain a positive tax characteristic corresponding to the positive tax data sample and a negative tax characteristic corresponding to the negative tax data sample;
constructing a comparative learning loss based on the difference between the training sample features and the positive example tax features and the difference between the training sample features and the negative example tax features;
and optimizing the feature extraction model to be trained based on the comparison learning loss to obtain the target feature extraction model.
Optionally, the tax risk prediction device is further configured to:
acquiring training tax associated data of each preset data type corresponding to a training user and a corresponding real label, and generating training tax characteristics corresponding to each training tax associated data;
based on a tax risk prediction model to be trained and each training tax characteristic, carrying out tax risk prediction on the training user to obtain an initial training prediction result;
performing model interpretation aiming at the tax risk prediction model to be trained on the initial training prediction result to evaluate the tampering degree of the tax-related data to obtain a training tampering degree evaluation result;
performing secondary tax risk prediction on the training user based on the training tampering degree evaluation result, the tax risk prediction model to be trained and each training tax characteristic to obtain a target training prediction result;
calculating a model total loss based on the real label, the initial training prediction result and the target training prediction result;
and optimizing the tax risk prediction model to be trained based on the model total loss to obtain the tax risk prediction model.
The specific implementation of the tax risk prediction device is basically the same as that of the embodiments of the tax risk prediction method, and is not described herein again.
The embodiment of the present application provides a readable storage medium, and the readable storage medium stores one or more programs, which can be further executed by one or more processors for implementing the steps of the tax risk prediction method according to any one of the above items.
The specific implementation manner of the readable storage medium of the present application is substantially the same as that of each embodiment of the tax risk prediction method, and is not described herein again.
The embodiments of the present application provide a computer program product, and the computer program product includes one or more computer programs, which can also be executed by one or more processors for implementing the steps of the tax risk prediction method of any one of the above.
The specific implementation of the computer program product of the present application is substantially the same as the embodiments of the tax risk prediction method, and is not further described herein.
The above description is only a preferred embodiment of the present application, and not intended to limit the scope of the present application, and all modifications of equivalent structures and equivalent processes, which are made by the contents of the specification and the drawings, or which are directly or indirectly applied to other related technical fields, are included in the scope of the present application.
Claims (10)
1. A tax risk prediction method, characterized by comprising:
acquiring a tax associated data set corresponding to a target user, and performing feature extraction on the tax associated data set based on a target feature extraction model to obtain a target tax feature set, wherein the feature extraction model is obtained by performing comparison learning based on preset tampered tax associated data and preset non-tampered tax associated data;
and performing tax risk prediction on the target user based on a tax risk prediction model and the target tax feature set to obtain a tax risk prediction result.
2. The tax risk prediction method of claim 1 wherein the step of predicting tax risk for the target user based on the tax risk prediction model and the target tax feature set to obtain a tax risk prediction result comprises:
based on the tax risk prediction model and the target tax characteristic set, carrying out tax risk prediction on the target user to obtain an initial tax risk prediction result;
performing model interpretation aiming at the tax risk prediction model on the initial tax risk prediction result to evaluate the tampering degree of each tax-related data to obtain a tampering degree evaluation result;
and performing secondary tax risk prediction on the target user based on the tampering degree evaluation result, the tax risk prediction model and the target tax feature set to obtain a target tax risk prediction result.
3. The tax risk prediction method of claim 2 wherein the set of target tax characteristics includes at least one target tax characteristic,
the step of performing secondary tax risk prediction on the target user based on the tampering degree evaluation result, the tax risk prediction model and the target tax feature set to obtain a target tax risk prediction result comprises:
generating target combination weights corresponding to the target tax characteristics based on the tampering degree evaluation result;
based on each target combination weight, performing weighted combination on each target tax characteristic to obtain a target combination tax characteristic;
and performing secondary tax risk prediction on the target user based on the target combined tax characteristics and the tax risk prediction model to obtain a target tax risk prediction result.
4. The tax risk prediction method of claim 2 wherein the tax-related data set comprises tax-related data of a predetermined data type, the target tax feature set comprises a target tax feature corresponding to the tax-related data,
the model interpretation of the initial tax risk prediction result for the tax risk prediction model is performed to evaluate the tampering degree of each tax-related data, and the step of obtaining the tampering degree evaluation result includes:
respectively calculating the characteristic contribution degree of each target tax characteristic to the initial tax risk prediction result based on the tax risk prediction model;
and evaluating the tampering degree of the tax related data based on the characteristic contribution degrees to obtain the tampering degree evaluation result.
5. The tax risk prediction method according to claim 4, wherein the step of evaluating the tampering degree of each tax-related data based on each of the feature contribution degrees, and obtaining the result of the evaluation of the tampering degree comprises:
based on a preset abnormity discrimination model, carrying out abnormity characteristic contribution degree discrimination on each characteristic contribution degree to obtain an abnormity discrimination result;
and evaluating the tampering degree of each tax-related data based on the abnormal judgment result to obtain a tampering degree evaluation result.
6. The tax risk prediction method according to claim 1, wherein before the step of performing feature extraction on the tax-related data set based on a target feature extraction model to obtain a target tax feature set, wherein the feature extraction model is obtained by performing comparative learning based on preset tampered tax-related data and preset untampered tax-related data, the tax risk prediction method further comprises:
acquiring a training sample and a sample tampering identifier corresponding to the training sample, and extracting a positive tax data sample and a negative tax data sample corresponding to the training sample from the preset tampering tax associated data and the preset non-tampering tax associated data based on the sample tampering identifier;
performing feature extraction on the training sample based on a feature extraction model to be trained to obtain training tax features corresponding to the training sample;
respectively extracting the characteristics of the positive tax data sample and the negative tax data sample based on a preset prior characteristic extraction model to obtain a positive tax characteristic corresponding to the positive tax data sample and a negative tax characteristic corresponding to the negative tax data sample;
constructing a comparative learning loss based on the difference between the training sample features and the positive example tax features and the difference between the training sample features and the negative example tax features;
and optimizing the feature extraction model to be trained based on the comparison learning loss to obtain the target feature extraction model.
7. The tax risk prediction method according to claim 1, wherein before the step of predicting tax risk to the target user based on the tax risk prediction model and the target tax feature set to obtain a tax risk prediction result, the tax risk prediction method further comprises:
acquiring training tax associated data of each preset data type corresponding to a training user and a corresponding real label, and generating training tax characteristics corresponding to each training tax associated data;
based on a tax risk prediction model to be trained and each training tax characteristic, carrying out tax risk prediction on the training user to obtain an initial training prediction result;
performing model interpretation aiming at the tax risk prediction model to be trained on the initial training prediction result to evaluate the tampering degree of the tax-related data to obtain a training tampering degree evaluation result;
performing secondary tax risk prediction on the training user based on the training tampering degree evaluation result, the tax risk prediction model to be trained and each training tax characteristic to obtain a target training prediction result;
calculating a model total loss based on the real label, the initial training prediction result and the target training prediction result;
and optimizing the tax risk prediction model to be trained based on the model total loss to obtain the tax risk prediction model.
8. A tax risk prediction device, characterized in that the tax risk prediction device comprises: a memory, a processor, and a program stored on the memory for implementing the tax risk prediction method,
the memory is used for storing a program for realizing the tax risk prediction method;
the processor is configured to execute a program implementing the tax risk prediction method to implement the steps of the tax risk prediction method according to any one of claims 1 to 7.
9. A medium which is a readable storage medium, characterized in that the readable storage medium has stored thereon a program for implementing a tax risk prediction method, the program being executed by a processor to implement the steps of the tax risk prediction method according to any one of claims 1 to 7.
10. A computer program product comprising a computer program, wherein the computer program, when executed by a processor, implements the steps of the method for object detection network construction optimization according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110553379.9A CN113269433B (en) | 2021-05-20 | 2021-05-20 | Tax risk prediction method, apparatus, medium and computer program product |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110553379.9A CN113269433B (en) | 2021-05-20 | 2021-05-20 | Tax risk prediction method, apparatus, medium and computer program product |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113269433A true CN113269433A (en) | 2021-08-17 |
| CN113269433B CN113269433B (en) | 2023-12-12 |
Family
ID=77232094
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110553379.9A Active CN113269433B (en) | 2021-05-20 | 2021-05-20 | Tax risk prediction method, apparatus, medium and computer program product |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113269433B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116452007A (en) * | 2023-06-15 | 2023-07-18 | 深圳市迪博企业风险管理技术有限公司 | Enterprise tax compliance risk assessment method based on capsule network |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108647818A (en) * | 2018-05-08 | 2018-10-12 | 山东浪潮通软信息科技有限公司 | A kind of method and device of prediction enterprise concerning taxes risk |
| CN108961243A (en) * | 2018-07-05 | 2018-12-07 | 广东工业大学 | It is a kind of that video labeling method and device are distorted based on machine learning |
| CN110097369A (en) * | 2019-03-18 | 2019-08-06 | 深圳壹账通智能科技有限公司 | Transaction data processing method, device, electronic equipment and storage medium |
| CN110930250A (en) * | 2020-02-12 | 2020-03-27 | 成都数联铭品科技有限公司 | Enterprise credit risk prediction method and system, storage medium and electronic equipment |
| CN111488623A (en) * | 2019-01-25 | 2020-08-04 | 深信服科技股份有限公司 | Webpage tampering detection method and related device |
| CN112785002A (en) * | 2021-03-15 | 2021-05-11 | 深圳前海微众银行股份有限公司 | Model construction optimization method, device, medium, and computer program product |
-
2021
- 2021-05-20 CN CN202110553379.9A patent/CN113269433B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108647818A (en) * | 2018-05-08 | 2018-10-12 | 山东浪潮通软信息科技有限公司 | A kind of method and device of prediction enterprise concerning taxes risk |
| CN108961243A (en) * | 2018-07-05 | 2018-12-07 | 广东工业大学 | It is a kind of that video labeling method and device are distorted based on machine learning |
| CN111488623A (en) * | 2019-01-25 | 2020-08-04 | 深信服科技股份有限公司 | Webpage tampering detection method and related device |
| CN110097369A (en) * | 2019-03-18 | 2019-08-06 | 深圳壹账通智能科技有限公司 | Transaction data processing method, device, electronic equipment and storage medium |
| CN110930250A (en) * | 2020-02-12 | 2020-03-27 | 成都数联铭品科技有限公司 | Enterprise credit risk prediction method and system, storage medium and electronic equipment |
| CN112785002A (en) * | 2021-03-15 | 2021-05-11 | 深圳前海微众银行股份有限公司 | Model construction optimization method, device, medium, and computer program product |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116452007A (en) * | 2023-06-15 | 2023-07-18 | 深圳市迪博企业风险管理技术有限公司 | Enterprise tax compliance risk assessment method based on capsule network |
| CN116452007B (en) * | 2023-06-15 | 2023-09-19 | 深圳市迪博企业风险管理技术有限公司 | Enterprise tax compliance risk assessment method based on capsule network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113269433B (en) | 2023-12-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10755196B2 (en) | Determining retraining of predictive models | |
| CN112801773A (en) | Enterprise risk early warning method, device, equipment and storage medium | |
| US20210192376A1 (en) | Automated, progressive explanations of machine learning results | |
| CN113837596B (en) | Fault determination method and device, electronic equipment and storage medium | |
| CN110033284A (en) | Source of houses verification method, apparatus, equipment and storage medium | |
| CN113269359B (en) | User financial status prediction method, device, medium, and computer program product | |
| CN111145006A (en) | Automobile financial anti-fraud model training method and device based on user portrait | |
| CN110688536A (en) | Label prediction method, device, equipment and storage medium | |
| CN113205403A (en) | Method and device for calculating enterprise credit level, storage medium and terminal | |
| CN112328869A (en) | User loan willingness prediction method and device and computer system | |
| CN112529575A (en) | Risk early warning method, equipment, storage medium and device | |
| CN113269433B (en) | Tax risk prediction method, apparatus, medium and computer program product | |
| Ouyang et al. | Modeling and forecasting the popularity evolution of mobile apps: A multivariate hawkes process approach | |
| CN111784053A (en) | Transaction risk detection method, device and readable storage medium | |
| CN112801489A (en) | Litigation case risk detection method, litigation case risk detection device, litigation case risk detection equipment and readable storage medium | |
| CN111582313B (en) | Sample data generation method and device and electronic equipment | |
| CN111815435A (en) | Visualization method, device, equipment and storage medium for group risk characteristics | |
| CN111582647A (en) | User data processing method and device and electronic equipment | |
| CN114387089A (en) | Client credit risk assessment method, device, equipment and storage medium | |
| CN110472680B (en) | Object classification method, device and computer-readable storage medium | |
| CN114416422A (en) | Problem locating method, apparatus, device, medium and program product | |
| CN114066513A (en) | User classification method and device | |
| CN112712423A (en) | Suspected illegal fundraising item judgment method and device, computer equipment and storage medium | |
| CN113807391A (en) | Task model training method and device, electronic equipment and storage medium | |
| CN113221995A (en) | Data classification method, equipment and device based on semi-supervised deep classification algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |