A declaration sample output method, device and medium
By extracting business matter samples from real declaration samples, learning calculation models and building indicator models, the problems of cumbersome filling of forms and complex calculation formulas in tax tax returns are solved, and higher declaration accuracy and flexibility are achieved.
Patent Information
- Application Number
- CN202410040888.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-01-10
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2044-01-10
AI Technical Summary
During the tax tax filing process, filling out forms is complicated and complicated, and the complexity of calculation formulas leads to errors and omissions, which increases the burden and troubles of taxpayers, especially for taxpayers who do not have professional knowledge.
By extracting business matter samples from multiple real declaration samples, learning calculation models, building business matter indicator models, and using pre-acquisition tax calculation methods, adapting to the situation of different tax types and taxpayers, simplifying the filling process, and improving calculation accuracy and flexibility.
The process of filling out tax returns is simplified, the accuracy and flexibility of calculations are improved, the dependence on professional knowledge is reduced, the risk of filling out errors and omissions is reduced, and the declaration efficiency is improved.
Smart Images

Figure CN117876134B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of computer technology, and in particular to a method, device and medium for outputting a declaration sample. Background Art
[0002] Tax declaration is an important task for enterprises and individuals to declare taxes to tax authorities in accordance with national tax laws. At present, tax declaration requires the use of inter-table and intra-table calculation formulas to calculate tax declaration data. Specifically, taxpayers fill in relevant information according to the forms specified by the tax authorities, and then calculate tax declaration data through inter-table and intra-table calculation formulas.
[0003] However, there are some problems with current tax returns. First, filling out forms is cumbersome and complicated, requiring taxpayers to spend a lot of time and energy on calculations and verification. This not only increases the burden on taxpayers, but also easily leads to errors and omissions in filling out forms, causing unnecessary troubles for taxpayers and tax authorities. Secondly, the complexity of calculation formulas between and within tables also causes trouble for taxpayers. Due to the different situations of different taxes and different taxpayers, the calculation formulas between and within tables need to be adjusted and applied according to the specific circumstances, which is a difficult problem for taxpayers who do not have professional knowledge. Summary of the invention
[0004] One or more embodiments of the present specification provide a method, device and medium for outputting a declaration sample, which are used to solve the technical problems raised by the background technology.
[0005] One or more embodiments of this specification adopt the following technical solutions:
[0006] One or more embodiments of this specification provide a method for outputting a declaration sample, including:
[0007] Extract samples of various business items from multiple real declaration samples;
[0008] According to the calculation methods of various tax types obtained in advance, the business item samples are learned, and the calculation models of the business item samples are respectively determined, so as to realize the reuse of the business items of the real declaration samples by applying the common features in the learned calculation models to the specific business items of the actual samples;
[0009] Constructing a business item indicator model according to the calculation model of each business item sample;
[0010] According to the business item indicator model and the pre-generated tax calculation rules, the business items currently being reported are calculated and the current reporting sample is output.
[0011] It should be noted that the above contents of the embodiments of this specification have the following beneficial effects:
[0012] Simplify the filling process: By extracting samples of various business items from multiple real declaration samples and learning the calculation models of each business item sample, taxpayers can reuse the business items of the real declaration samples by applying the common features in the learned calculation models to specific business items of the actual samples. In this way, taxpayers can complete the calculation and verification more simply and quickly when filling in the declaration form, reducing the tediousness and complexity of filling in.
[0013] Improve accuracy: By learning the calculation model of each business item sample and building a business item indicator model, the declared data can be calculated more accurately. For taxpayers without professional knowledge, the complexity of the calculation formulas between and within the tables may lead to errors and omissions, and this method can reduce the occurrence of these problems and improve the accuracy of tax declarations.
[0014] Increase calculation flexibility: Obtain the calculation methods of various taxes in advance and apply them to the calculation models of different business items, which can flexibly adapt to different taxes and different taxpayers. In this way, taxpayers no longer need to deal with the adjustment and application of calculation formulas between tables and within tables separately, which reduces the dependence on professional knowledge and improves the flexibility of calculation.
[0015] In summary, this method of outputting declaration samples can simplify the filling process, improve accuracy, and increase calculation flexibility.
[0016] Furthermore, the extraction of each business item sample from multiple real declaration samples described in one or more embodiments of this specification includes:
[0017] Calculate the entropy value and information gain of the features in the multiple real declaration samples respectively, and match the optimal feature combination;
[0018] Segmenting the multiple real declaration samples according to the optimal feature combination to form an accurate decision tree;
[0019] According to the accurate decision tree, samples of each business item are extracted.
[0020] It should be noted that the above method of extracting samples of various business items from multiple real declaration samples has the following specific beneficial effects:
[0021] Improve the accuracy of sample extraction: By calculating the entropy value and analyzing the information gain of the features in multiple real declaration samples, the optimal feature combination can be matched to improve the accuracy of sample extraction. In this way, the extracted business item samples can more truly reflect the characteristics and rules of the real declaration data, providing a more accurate data basis for the subsequent calculation model construction.
[0022] Furthermore, one or more embodiments of this specification respectively calculate entropy values and information gains for the features in the multiple real declaration samples to match the optimal feature combination, including:
[0023] For the features in the multiple real declaration samples, respectively calculate the entropy value of each feature, where the entropy value is used to measure the degree of confusion or uncertainty of the feature in the sample;
[0024] For the features in the multiple real declaration samples, the information gain of each feature is calculated by the entropy value of each feature, and the information gain measures the degree to which the uncertainty of the result is reduced under given feature conditions;
[0025] According to the information gain of each feature, the feature with the maximum information gain is selected as the optimal feature combination.
[0026] It should be noted that the method of calculating entropy and information gain of features in multiple real declaration samples and matching the optimal feature combination has the following specific beneficial effects:
[0027] Determine the optimal feature combination: By calculating the entropy and information gain of the features in multiple real declaration samples, we can evaluate the degree of confusion of the features in the samples and the degree of influence of the features on the uncertainty of the results. By selecting the features with the largest information gain, we can determine the optimal feature combination, that is, the features that have the greatest impact on the results.
[0028] Improve the accuracy of feature selection: By calculating entropy and information gain, we can objectively evaluate the impact of each feature on the result. Selecting the feature with the largest information gain as the optimal feature combination can more accurately select the features that have the greatest impact on the result, providing an accurate basis for subsequent sample segmentation and decision tree construction.
[0029] Optimize sample segmentation and decision tree construction: By selecting the optimal feature combination, multiple real declaration samples can be segmented to form an accurate decision tree. An accurate decision tree can better reflect the characteristics and rules of real declaration data, and provide a more accurate and reliable basis for subsequent calculation model construction and business case sample extraction.
[0030] In summary, this method of calculating entropy and information gain for features in multiple real declaration samples and matching the optimal feature combination can determine the optimal feature combination, improve the accuracy of feature selection, and optimize sample segmentation and decision tree construction.
[0031] Furthermore, one or more embodiments of this specification describe segmenting the multiple real declaration samples according to the optimal feature combination to form an accurate decision tree, including:
[0032] Perform the first segmentation of multiple real declaration samples according to the optimal feature combination to form an initial decision tree;
[0033] Cyclic segmentation of multiple real declared samples according to the optimal feature combination, and determining whether to continue segmentation according to predefined conditions, wherein the predefined conditions include a predefined maximum depth or a predefined number of samples;
[0034] If the predefined conditions are met, the accurate decision tree is formed.
[0035] It should be noted that the above-mentioned formation of an accurate decision tree has the following specific beneficial effects:
[0036] Improve the accuracy of decision trees: By using the optimal feature combination to segment multiple real declaration samples, it can be ensured that the nodes and branches of the decision tree can accurately capture the characteristics and patterns of the real declaration data. In this way, the accurate decision tree formed can more accurately predict the business item data of the current declaration sample and reduce the possibility of filling errors and omissions.
[0037] Simplify the decision tree construction process: By specifying predefined conditions, such as a predefined maximum depth or number of samples, the complexity and size of the decision tree can be controlled. This simplifies the decision tree construction process, reduces unnecessary splits and nodes, and improves the interpretability and usability of the decision tree.
[0038] Improve segmentation efficiency: By using the optimal feature combination for cyclic segmentation and judging whether to continue segmentation based on predefined conditions, the efficiency of segmentation can be improved. The next round of segmentation is performed only when the predefined conditions are met, avoiding unnecessary calculations and processing, saving time and resources.
[0039] In summary, segmenting multiple real declaration samples according to the optimal feature combination to form an accurate decision tree can improve the accuracy of the decision tree, simplify the decision tree construction process, and improve the segmentation efficiency.
[0040] Furthermore, in one or more embodiments of this specification, extracting the samples of each business item according to the accurate decision tree includes:
[0041] Inputting multiple real declaration samples into an accurate decision tree to determine the business item to which each real declaration sample belongs according to the path of each real declaration sample on the accurate decision tree;
[0042] Each real declaration sample is extracted according to the business item to which it belongs, so as to extract the real declaration samples belonging to a specific business item from a plurality of real declaration sample sets to form samples of each business item.
[0043] It should be noted that the specific beneficial effects of the above content are as follows:
[0044] Automatic classification: Classifying multiple real declaration samples can automatically identify different business matters without human intervention, reducing the risk of human errors and improving the accuracy and efficiency of classification.
[0045] Accurately extract samples: Through accurate decision tree path judgment, the specific business items to which each real declaration sample belongs can be determined, and then the real declaration samples belonging to specific business items can be extracted from multiple real declaration samples. This can reduce the tedious steps and verification work of taxpayers when filling out forms, and improve the accuracy and efficiency of filling out forms.
[0046] Optimize calculation model: By extracting samples from accurate decision trees, the calculation model of each business item sample can be determined more accurately. This allows taxpayers to perform operations based on pre-determined calculation models when performing calculations, reducing the trouble of complex calculation formulas between tables and within tables, and increasing the accuracy and convenience of operations.
[0047] Furthermore, one or more embodiments of this specification describe that, according to the calculation methods of various tax types obtained in advance, the business item samples are learned, and the calculation models of the business item samples are respectively determined, so as to realize the reuse of the business items of the actual declaration samples by applying the common features in the learned calculation models to the specific business items of the actual samples, including:
[0048] Extracting each business item sample from the multiple real declaration samples;
[0049] Based on the previously acquired tax calculation methods, the samples of each business item are studied to extract the common features of the samples;
[0050] The calculation model of each business item sample is determined based on the common characteristics of the samples and the calculation methods of each tax type, so that the common characteristics in the calculation model of each business item sample can be applied to the specific business items of the actual sample to achieve the reuse of the business items of the actual declaration sample.
[0051] It should be noted that the specific beneficial effects of the above content are as follows:
[0052] Extracting common features of samples: Based on the pre-acquired tax calculation methods, the business item samples are studied to extract common features between samples. These common features can be used to build a calculation model to achieve reuse of sample business items.
[0053] Constructing calculation models: Based on the common characteristics of the samples and the calculation methods of various taxes, the calculation model of each business item sample can be determined. In this way, taxpayers can operate based on the predetermined calculation model when filing, reducing the tedious calculation and verification workload when filling out forms, and improving the accuracy and efficiency of filing.
[0054] To summarize, by learning and determining the calculation model for each business item sample based on the calculation method of each tax type obtained in advance, and applying common features to achieve the reuse of business items of real declaration samples, the common features of the samples can be extracted and the calculation model can be constructed, thereby reducing the tediousness of filling out forms and improving the accuracy and efficiency of declarations.
[0055] Furthermore, one or more embodiments of this specification calculate the business items currently reported based on the business item indicator model and the pre-generated tax calculation rules, and output the current report sample, including:
[0056] Comparing the current declaration sample with the multiple real declaration samples to determine similar business items of the current declaration sample;
[0057] According to the business item indicator model and the pre-generated tax calculation rules, the similar business items are calculated and the current declaration sample is output.
[0058] It should be noted that the specific beneficial effects of the above content are as follows:
[0059] Determine similar business items: By comparing the current declaration sample with multiple real declaration samples, you can determine which real declaration samples have similar business items with the current declaration sample. This allows you to more accurately calculate the current declaration sample based on similar business items.
[0060] Improve calculation accuracy: Calculating similar business items based on the business item indicator model and pre-generated tax calculation rules can improve calculation accuracy. By using pre-determined calculation rules, it reduces the trouble of complex calculation formulas between tables and within tables, and reduces the possibility of filling errors and omissions.
[0061] Improve calculation efficiency: By calculating similar business items based on the business item indicator model and pre-generated tax calculation rules, the calculation efficiency can be improved. This reduces the time and energy taxpayers spend on complex calculations and verification when filling out forms, and improves the efficiency of declaration.
[0062] In summary, according to the business item indicator model and the pre-generated tax calculation rules, the currently declared business items are calculated, and the output of the current declared sample can determine similar business items, improve the calculation accuracy and efficiency, thereby solving the problems of cumbersome and complicated form filling and the trouble of calculation formulas between and within tables.
[0063] Furthermore, after outputting the current application sample in one or more embodiments of this specification, the method further includes:
[0064] Analyzing difference parameters according to the comparison results of the current declared sample and the multiple real declared samples;
[0065] The tax calculation rules are optimized and adjusted according to the difference parameters.
[0066] It should be noted that the specific beneficial effects of the above content are as follows:
[0067] Refined tax calculation rules: By comparing and analyzing multiple real declaration samples, the difference parameters between the current declaration sample and the real declaration sample can be determined. These difference parameters reflect the specific differences between the current declaration sample and the real declaration sample in the calculation process. By analyzing the comparison results, we can deeply understand the differences and make more refined optimization and adjustment of the tax calculation rules.
[0068] Improve calculation accuracy and adaptability: Optimize and adjust the tax calculation rules according to the difference parameters, and customize the calculation according to the specific situation of the current declaration sample. By optimizing and adjusting the tax calculation rules, various taxes and fees related to the current declaration sample can be calculated more accurately, improving the accuracy and adaptability of the calculation.
[0069] Reduce filling errors and omissions: By optimizing and adjusting the tax calculation rules, it can be made more consistent with the characteristics and regulations of the current declaration sample, reducing the occurrence of filling errors and omissions. Adaptable calculation rules can reduce taxpayers' troubles with complex inter-table and intra-table calculation formulas, reduce the risk of errors and omissions when filling in forms, and reduce the burden on taxpayers.
[0070] One or more embodiments of this specification provide a declaration sample output device, including:
[0071] at least one processor; and,
[0072] a memory communicatively connected to the at least one processor; wherein,
[0073] The memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to:
[0074] Extract samples of various business items from multiple real declaration samples;
[0075] According to the calculation methods of various tax types obtained in advance, the business item samples are learned, and the calculation models of the business item samples are determined respectively, so as to realize the reuse of the business items of the real declaration samples by applying the common features in the learned calculation models to the specific business items of the actual samples;
[0076] Constructing a business item indicator model according to the calculation model of each business item sample;
[0077] According to the business item indicator model and the pre-generated tax calculation rules, the business items currently being reported are calculated and the current reporting sample is output.
[0078] One or more embodiments of this specification provide a non-volatile computer storage medium storing computer executable instructions, which can achieve the following when executed by a computer:
[0079] Extract samples of various business items from multiple real declaration samples;
[0080] According to the calculation methods of various tax types obtained in advance, the business item samples are learned, and the calculation models of the business item samples are determined respectively, so as to realize the reuse of the business items of the real declaration samples by applying the common features in the learned calculation models to the specific business items of the actual samples;
[0081] Constructing a business item indicator model according to the calculation model of each business item sample;
[0082] According to the business item indicator model and the pre-generated tax calculation rules, the business items currently being reported are calculated and the current reporting sample is output.
[0083] At least one of the above technical solutions adopted in the embodiments of this specification can achieve the following beneficial effects:
[0084] Simplify the filling process: By extracting samples of various business items from multiple real declaration samples and learning the calculation models of each business item sample, taxpayers can reuse the business items of the real declaration samples by applying the common features in the learned calculation models to specific business items of the actual samples. In this way, taxpayers can complete the calculation and verification more simply and quickly when filling in the declaration form, reducing the tediousness and complexity of filling in.
[0085] Improve accuracy: By learning the calculation model of each business item sample and building a business item indicator model, the declared data can be calculated more accurately. For taxpayers without professional knowledge, the complexity of the calculation formulas between and within the tables may lead to errors and omissions, and this method can reduce the occurrence of these problems and improve the accuracy of tax declarations.
[0086] Increase calculation flexibility: Obtain the calculation methods of various taxes in advance and apply them to the calculation models of different business items, which can flexibly adapt to different taxes and different taxpayers. In this way, taxpayers no longer need to deal with the adjustment and application of calculation formulas between tables and within tables separately, which reduces the dependence on professional knowledge and improves the flexibility of calculation.
[0087] In summary, this method of outputting declaration samples can simplify the filling process, improve accuracy, and increase calculation flexibility. BRIEF DESCRIPTION OF THE DRAWINGS
[0088] In order to more clearly illustrate the technical solutions in the embodiments of this specification or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art description. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. In the drawings:
[0089] Figure 1 A flowchart of a method for outputting a declaration sample provided for one or more embodiments of this specification;
[0090] Figure 2 A schematic diagram of an intelligent tax calculation engine based on an artificial neural network provided for one or more embodiments of this specification;
[0091] Figure 3 A schematic diagram of the structure of a declaration sample output device provided for one or more embodiments of this specification. DETAILED DESCRIPTION
[0092] The embodiments of this specification provide a method, device and medium for outputting a declaration sample.
[0093] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.
[0094] Figure 1 A schematic diagram of a process flow of a declaration sample output method provided in one or more embodiments of this specification, which process can be executed by a declaration sample output system. Certain input parameters or intermediate results in the process allow manual intervention and adjustment to help improve accuracy.
[0095] The method steps of the embodiment of this specification are as follows:
[0096] S102, extracting samples of various business items from a plurality of real declaration samples.
[0097] In an embodiment of the present specification, the entropy value and information gain of the features in the multiple real declaration samples can be calculated respectively to match the optimal feature combination; then, the multiple real declaration samples can be segmented according to the optimal feature combination to form an accurate decision tree; finally, according to the accurate decision tree, the samples of each business item can be extracted.
[0098] It should be noted that for multiple real declaration samples, the entropy and information gain can be calculated for the features in each sample. The entropy can be calculated by counting the frequency of occurrence of each feature value and then using the entropy formula to calculate the entropy. The information gain can be calculated by calculating the conditional entropy of each feature on the target variable and then subtracting the entropy of the target variable. According to the calculation results, the optimal feature combination is matched. The criterion for selecting the feature combination can be the feature combination with the maximum information gain or the minimum entropy.
[0099] The optimal feature combination is then used to segment multiple real application samples to form an accurate decision tree. The decision tree can be constructed using a recursive segmentation method, starting from the root node, classifying the samples according to the optimal feature combination, and then repeating the process for each child node until all samples are correctly classified or the stopping condition is reached.
[0100] Finally, according to the accurate decision tree, the samples of each business item are extracted. According to the structure and characteristic conditions of the decision tree, the new declaration samples are predicted, and the corresponding business item samples are extracted according to the prediction results.
[0101] It should be noted that the above method of extracting samples of various business items from multiple real declaration samples has the following specific beneficial effects:
[0102] Improve the accuracy of sample extraction: By calculating the entropy value and analyzing the information gain of the features in multiple real declaration samples, the optimal feature combination can be matched to improve the accuracy of sample extraction. In this way, the extracted business item samples can more truly reflect the characteristics and rules of the real declaration data, providing a more accurate data basis for the subsequent calculation model construction.
[0103] Furthermore, one or more embodiments of the present specification calculate entropy values and information gains for the features in the multiple real declaration samples respectively. When matching the optimal feature combination, the entropy value of each feature in the multiple real declaration samples can be calculated respectively, and the entropy value is used to measure the degree of confusion or uncertainty of the feature in the sample; for the features in the multiple real declaration samples, the information gain of each feature is calculated by the entropy value of each feature, and the information gain measures the degree to which the uncertainty of the result is reduced under given feature conditions; according to the information gain of each feature, the feature with the maximum information gain is selected as the optimal feature combination.
[0104] It should be noted that the above-mentioned matching of the optimal feature combination can be carried out through the following specific steps:
[0105] For multiple real declaration samples, the entropy value is calculated for each feature. The entropy value can be calculated by counting the frequency of occurrence of each feature value and then using the entropy formula. The entropy value is used to measure the degree of confusion or uncertainty of the feature in the sample. For multiple real declaration samples, the information gain of each feature can be calculated by the entropy value of each feature. Information gain can be calculated by calculating the conditional entropy of each feature on the target variable and then subtracting the entropy value of the target variable. Information gain measures the degree to which the uncertainty of the result is reduced under given feature conditions. According to the information gain of each feature, the feature with the maximum information gain is selected as the optimal feature combination. The feature with the maximum information gain indicates that under this feature condition, it has the greatest impact on the classification of the target variable, that is, it can best reduce the uncertainty of the result.
[0106] It should be noted that the method of calculating entropy and information gain of features in multiple real declaration samples and matching the optimal feature combination has the following specific beneficial effects:
[0107] Determine the optimal feature combination: By calculating the entropy and information gain of the features in multiple real declaration samples, we can evaluate the degree of confusion of the features in the samples and the degree of influence of the features on the uncertainty of the results. By selecting the features with the largest information gain, we can determine the optimal feature combination, that is, the features that have the greatest impact on the results.
[0108] Improve the accuracy of feature selection: By calculating entropy and information gain, we can objectively evaluate the impact of each feature on the result. Selecting the feature with the largest information gain as the optimal feature combination can more accurately select the features that have the greatest impact on the result, providing an accurate basis for subsequent sample segmentation and decision tree construction.
[0109] Optimize sample segmentation and decision tree construction: By selecting the optimal feature combination, multiple real declaration samples can be segmented to form an accurate decision tree. An accurate decision tree can better reflect the characteristics and rules of real declaration data, and provide a more accurate and reliable basis for subsequent calculation model construction and business case sample extraction.
[0110] In summary, this method of calculating entropy and information gain for features in multiple real declaration samples and matching the optimal feature combination can determine the optimal feature combination, improve the accuracy of feature selection, and optimize sample segmentation and decision tree construction.
[0111] Furthermore, when the multiple real reported samples are segmented according to the optimal feature combination to form an accurate decision tree as described in one or more embodiments of the present specification, the multiple real reported samples can be first segmented according to the optimal feature combination to form an initial decision tree; then the multiple real reported samples can be segmented cyclically according to the optimal feature combination, and whether to continue segmenting can be determined according to predefined conditions, wherein the predefined conditions include a predefined maximum depth or a predefined number of samples; if the predefined conditions are met, the accurate decision tree is formed.
[0112] It should be noted that the above-mentioned accurate decision tree can be formed through the following specific steps:
[0113] According to the optimal feature combination, multiple real declaration samples are first segmented to form an initial decision tree. The samples can be divided according to the optimal feature combination, and samples with the same feature value are classified into the same sub-node.
[0114] Then, the segmentation is repeated according to the optimal feature combination, and the decision of whether to continue segmentation is made according to the predefined conditions. The predefined conditions can be the maximum depth or the number of samples. If the predefined conditions are met, the segmentation continues, otherwise the segmentation stops.
[0115] In the process of cyclic segmentation, a recursive algorithm can be used to further segment each sub-node. In each segmentation, the feature with the maximum information gain is selected as the segmentation basis according to the optimal feature combination, and the sample is further divided into smaller sub-nodes.
[0116] When the predefined conditions are met, the loop segmentation stops and the final accurate decision tree is formed. Each leaf node of the accurate decision tree represents a business item sample, and the sample on the leaf node is identified as belonging to the corresponding business item.
[0117] It should be noted that the above-mentioned formation of an accurate decision tree has the following specific beneficial effects:
[0118] Improve the accuracy of decision trees: By using the optimal feature combination to segment multiple real declaration samples, it can be ensured that the nodes and branches of the decision tree can accurately capture the characteristics and patterns of the real declaration data. In this way, the accurate decision tree formed can more accurately predict the business item data of the current declaration sample and reduce the possibility of filling errors and omissions.
[0119] Simplify the decision tree construction process: By specifying predefined conditions, such as a predefined maximum depth or number of samples, the complexity and size of the decision tree can be controlled. This simplifies the decision tree construction process, reduces unnecessary splits and nodes, and improves the interpretability and usability of the decision tree.
[0120] Improve segmentation efficiency: By using the optimal feature combination for cyclic segmentation and judging whether to continue segmentation based on predefined conditions, the efficiency of segmentation can be improved. The next round of segmentation is performed only when the predefined conditions are met, avoiding unnecessary calculations and processing, saving time and resources.
[0121] In summary, segmenting multiple real declaration samples according to the optimal feature combination to form an accurate decision tree can improve the accuracy of the decision tree, simplify the decision tree construction process, and improve the segmentation efficiency.
[0122] Furthermore, when extracting the samples of each business matter according to the accurate decision tree as described in one or more embodiments of the present specification, multiple real declaration samples can be input into the accurate decision tree to determine the business matter to which each real declaration sample belongs according to the path of each real declaration sample on the accurate decision tree; each real declaration sample is extracted according to the business matter to which each real declaration sample belongs, so as to extract the real declaration samples belonging to a specific business matter from the multiple real declaration sample sets to form each business matter sample.
[0123] It should be noted that the above-mentioned extraction of the business item samples can be carried out through the following specific implementation steps:
[0124] Input multiple real declaration samples into the accurate decision tree. For each real declaration sample, you can start from the root node and traverse downward step by step according to the judgment conditions of the accurate decision tree, and determine the business item to which each sample belongs based on the path of each sample on the decision tree. According to the business item to which each real declaration sample belongs, extract it from multiple real declaration sample sets. Classify the real declaration samples belonging to a specific business item into the corresponding business item sample set.
[0125] When forming samples of each business item, each real declaration sample can be extracted from multiple real declaration sample sets according to the business item to which it belongs, so as to form a sample set of each business item.
[0126] It should be noted that the specific beneficial effects of the above content are as follows:
[0127] Automatic classification: Classifying multiple real declaration samples can automatically identify different business matters without human intervention, reducing the risk of human errors and improving the accuracy and efficiency of classification.
[0128] Accurately extract samples: Through accurate decision tree path judgment, the specific business items to which each real declaration sample belongs can be determined, and then the real declaration samples belonging to specific business items can be extracted from multiple real declaration samples. This can reduce the tedious steps and verification work of taxpayers when filling out forms, and improve the accuracy and efficiency of filling out forms.
[0129] Optimize calculation model: By extracting samples from accurate decision trees, the calculation model of each business item sample can be determined more accurately. This allows taxpayers to perform operations based on pre-determined calculation models when performing calculations, reducing the trouble of complex calculation formulas between tables and within tables, and increasing the accuracy and convenience of operations.
[0130] S104, based on the calculation method of each tax type obtained in advance, the business item samples are studied, and the calculation models of the business item samples are respectively determined, so as to realize the reuse of the business items of the actual declaration samples by applying the common features in the learned calculation model to the specific business items of the actual samples.
[0131] In the embodiment of the present specification, each business item sample can be first extracted from the multiple real declaration samples, and the multiple real declaration samples can be classified according to the business items according to the existing business item classification standards to extract the sample set belonging to each business item; then, according to the pre-acquired tax calculation methods, the business item samples are studied to extract the common features of the samples, and the common features related to the tax calculation methods can be obtained by analyzing the features of the business item samples. These common features can be features that can be shared by multiple business items, such as income, cost, profit, etc.; finally, the calculation model of each business item sample is determined according to the common features of the samples and the tax calculation methods, so that the common features in the calculation model of each business item sample can be applied to the specific business items of the actual samples to realize the reuse of the business items of the real declaration samples. The calculation model of each business item sample can be determined according to the common features of the samples and the tax calculation methods. According to the extracted common features, combined with the tax calculation methods, a calculation model for each business item is established. The computational model should be able to apply common features to specific business items of actual samples to achieve the reuse of business items of real declaration samples.
[0132] It should be noted that the specific beneficial effects of the above content are as follows:
[0133] Extracting common features of samples: Based on the pre-acquired tax calculation methods, the business item samples are studied to extract common features between samples. These common features can be used to build a calculation model to achieve reuse of sample business items.
[0134] Constructing calculation models: Based on the common characteristics of the samples and the calculation methods of various taxes, the calculation model of each business item sample can be determined. In this way, taxpayers can operate based on the predetermined calculation model when filing, reducing the tedious calculation and verification workload when filling out forms, and improving the accuracy and efficiency of filing.
[0135] To summarize, by learning and determining the calculation model for each business item sample based on the calculation method of each tax type obtained in advance, and applying common features to achieve the reuse of business items of real declaration samples, the common features of the samples can be extracted and the calculation model can be constructed, thereby reducing the tediousness of filling out forms and improving the accuracy and efficiency of declarations.
[0136] S106: construct a business item indicator model according to the calculation model of each business item sample.
[0137] In the embodiment of this specification, the above-mentioned construction of the business item indicator model can be carried out through the following specific steps:
[0138] According to the calculation model of each business item sample, determine to use the neural network model to construct the business item indicators. The neural network model can be used to model complex nonlinear relationships and can extract features and make predictions by learning a large amount of sample data. Collect and organize training data sets related to the construction of business item indicators. The training data set should include input variables and corresponding target variables (business item indicators). Design the structure of the neural network model. Determine the number of layers of the neural network, the number of neurons in each layer, the activation function, etc., and make corresponding adjustments based on the characteristics of the business item.
[0139] Train the neural network. Input the training data set into the neural network model, train the model through the back propagation algorithm, and optimize the weights and biases of the model so that it can better fit the training data set.
[0140] Validate and tune the model. Use the validation dataset to evaluate the performance of the model, and adjust and optimize the model to improve its generalization ability and accuracy.
[0141] Use the trained and verified neural network model to predict business item indicators. Input new input data into the trained model and calculate the corresponding business item indicators through forward propagation. For each business item, build a business item indicator model based on the predicted indicator results. The model can include the structure, parameter settings and related background knowledge of the neural network model.
[0142] S108, calculating the business items currently being reported according to the business item indicator model and the pre-generated tax calculation rules, and outputting the current reported sample.
[0143] In an embodiment of the present specification, the current declaration sample can be compared with the multiple real declaration samples to determine similar business items of the current declaration sample; then, the similar business items can be calculated based on the business item indicator model and the pre-generated tax calculation rules to output the current declaration sample.
[0144] It should be noted that the above-mentioned output of the current declaration sample can be implemented through the following specific steps:
[0145] Collect relevant information of multiple real declaration samples and the current declaration sample, which may include business items and related data. Compare the current declaration sample with multiple real declaration samples, and determine similar business items of the current declaration sample through comparison. Data matching algorithms or machine learning algorithms may be used for comparison and similarity calculation.
[0146] Similar business items are calculated based on the business item indicator model and the pre-generated tax calculation rules. The business item indicator model can be a classification model for classifying business items into different categories. The tax calculation rules can be pre-set tax methods and calculation formulas for each category.
[0147] Output the calculation results of the current declaration sample: Based on the calculation results, the corresponding tax return form or report can be generated for the declarant's reference and submission.
[0148] Verify and review the calculation results: Confirm the accuracy and rationality of the calculation results by comparing them with other real declared samples, and make further adjustments and corrections if necessary.
[0149] Prepare and submit declaration materials: Generate corresponding declaration materials based on calculation results, including tax return form, appendices, relevant supporting documents, etc., and submit them to tax authorities in accordance with relevant regulations and procedures.
[0150] Monitor and track application results: Monitor application materials, and keep abreast of application results and related feedback information. If necessary, make adjustments and corrections in a timely manner to ensure the accuracy and compliance of the application.
[0151] Conduct regular data analysis and model updates: Based on actual reporting conditions and feedback information, regularly update and optimize business item indicator models and tax calculation rules to improve calculation accuracy and practicality.
[0152] It should be noted that the specific beneficial effects of the above content are as follows:
[0153] Determine similar business items: By comparing the current declaration sample with multiple real declaration samples, you can determine which real declaration samples have similar business items with the current declaration sample. This allows you to more accurately calculate the current declaration sample based on similar business items.
[0154] Improve calculation accuracy: Calculating similar business items based on the business item indicator model and pre-generated tax calculation rules can improve calculation accuracy. By using pre-determined calculation rules, it reduces the trouble of complex calculation formulas between tables and within tables, and reduces the possibility of filling errors and omissions.
[0155] Improve calculation efficiency: By calculating similar business items based on the business item indicator model and pre-generated tax calculation rules, the calculation efficiency can be improved. This reduces the time and energy taxpayers spend on complex calculations and verification when filling out forms, and improves the efficiency of declaration.
[0156] In summary, according to the business item indicator model and the pre-generated tax calculation rules, the currently declared business items are calculated, and the output of the current declared sample can determine similar business items, improve the calculation accuracy and efficiency, thereby solving the problems of cumbersome and complicated form filling and the trouble of calculation formulas between and within tables.
[0157] Furthermore, after outputting the current declaration sample as described in one or more embodiments of the present specification, the difference parameters can be analyzed based on the comparison results between the current declaration sample and the multiple real declaration samples; and then the tax calculation rules can be optimized and adjusted based on the difference parameters.
[0158] It should be noted that the specific beneficial effects of the above content are as follows:
[0159] Refined tax calculation rules: By comparing and analyzing multiple real declaration samples, the difference parameters between the current declaration sample and the real declaration sample can be determined. These difference parameters reflect the specific differences between the current declaration sample and the real declaration sample in the calculation process. By analyzing the comparison results, we can deeply understand the differences and make more refined optimization and adjustment of the tax calculation rules.
[0160] Improve calculation accuracy and adaptability: Optimize and adjust the tax calculation rules according to the difference parameters, and customize the calculation according to the specific situation of the current declaration sample. By optimizing and adjusting the tax calculation rules, various taxes and fees related to the current declaration sample can be calculated more accurately, improving the accuracy and adaptability of the calculation.
[0161] Reduce filling errors and omissions: By optimizing and adjusting the tax calculation rules, it can be made more consistent with the characteristics and regulations of the current declaration sample, reducing the occurrence of filling errors and omissions. Adaptable calculation rules can reduce taxpayers' troubles with complex inter-table and intra-table calculation formulas, reduce the risk of errors and omissions when filling in forms, and reduce the burden on taxpayers.
[0162] It should be noted that the above contents of the embodiments of this specification have the following beneficial effects:
[0163] Simplify the filling process: By extracting samples of various business items from multiple real declaration samples and learning the calculation models of each business item sample, taxpayers can reuse the business items of the real declaration samples by applying the common features in the learned calculation models to specific business items of the actual samples. In this way, taxpayers can complete the calculation and verification more simply and quickly when filling in the declaration form, reducing the tediousness and complexity of filling in.
[0164] Improve accuracy: By learning the calculation model of each business item sample and building a business item indicator model, the declared data can be calculated more accurately. For taxpayers without professional knowledge, the complexity of the calculation formulas between and within the tables may lead to errors and omissions, and this method can reduce the occurrence of these problems and improve the accuracy of tax declarations.
[0165] Increase calculation flexibility: Obtain the calculation methods of various taxes in advance and apply them to the calculation models of different business items, which can flexibly adapt to different taxes and different taxpayers. In this way, taxpayers no longer need to deal with the adjustment and application of calculation formulas between tables and within tables separately, which reduces the dependence on professional knowledge and improves the flexibility of calculation.
[0166] In summary, this method of outputting declaration samples can simplify the filling process, improve accuracy, and increase calculation flexibility.
[0167] It should be noted that Figure 2 A schematic diagram of an intelligent tax calculation engine based on an artificial neural network provided in an embodiment of this specification is divided into seven steps: initializing the calculation engine, extracting data set samples, extracting business item models, building indicator models, rule engine executing calculations, adjusting sample rules, and forming a test set.
[0168] (1) Initialize the calculation engine
[0169] Initialize the calculation engine according to the technical characteristics of tax type, period, etc. Load different tax calculation modules.
[0170] (2) Extracting dataset samples
[0171] According to the features x, y, z...n in the initial parameters of the real declared samples, the entropy value and information gain are calculated to match the optimal feature combination as the splitting criterion and continuously split the sample set. The splitting is performed repeatedly before the stop condition is met to form an accurate decision tree. Thus, multiple corresponding samples are matched.
[0172] For example, if the sample set contains multiple features such as organization, period, tax type, and declaration method in the experience declaration data, then the tax type feature is added to reduce the confusion of the sample, thereby finding declaration data with a high degree of matching from the experience declaration data through the feature combination of period + tax type.
[0173] (3) Extracting business item models
[0174] Extract (2) to match each business item in the experience declaration data, conduct multi-sample analysis on the attributes of the business items, and gradually determine the calculation model of each business item. Apply the common features learned in the model to match the business items of the actual sample to reuse the calculation mode of the data items of the historical sample.
[0175] For example: VAT tax returns require taxpayers to fill in their unified social credit code for each period, and the returns for each reporting period and each tax type are consistent, so there is no need to recalculate. The values of the historical reporting period can be automatically extracted according to different organizations. For input invoice data, it is necessary to automatically calculate the actual input invoice data of the current period based on the combination of various conditions such as organization, period, and tax type, combined with the calculation method of the historical reporting form. However, the different input invoices in the current reporting data can be determined and classified according to the calculation rules of the historical reporting data. And a more accurate classification calculation rule can be formed for more accurate input invoice classification calculation of later invoices.
[0176] (4) Constructing an indicator model
[0177] Learn the calculation models for each business item, supplement, correct and optimize the indicator model library model, and improve the generalization ability of the model.
[0178] (5) Rule engine performs calculations
[0179] According to the constructed business item indicator model, the current declaration calculation is performed through the tax calculation rules. The corresponding declaration data sample is output.
[0180] (6) Adjust sample rules
[0181] For the training set and validation set sample data, combined with the real declaration samples constructed this time, the indicator level is improved by the declaration data of the declaration sample directionality and hit rate, and the calculation complexity is reduced. The calculation rules of tax declaration are automatically optimized and adjusted.
[0182] (7) Forming a test set
[0183] The test set formed after the calculation engine is completed will feed back the tax indicator calculation samples, forming a richer and more accurate tax indicator data set.
[0184] Figure 3 A schematic diagram of a structure of a declaration sample output device provided for one or more embodiments of this specification includes:
[0185] at least one processor; and,
[0186] a memory communicatively connected to the at least one processor; wherein,
[0187] The memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to:
[0188] Extract samples of various business items from multiple real declaration samples;
[0189] According to the calculation methods of various tax types obtained in advance, the business item samples are learned, and the calculation models of the business item samples are respectively determined, so as to realize the reuse of the business items of the real declaration samples by applying the common features in the learned calculation models to the specific business items of the actual samples;
[0190] Constructing a business item indicator model according to the calculation model of each business item sample;
[0191] According to the business item indicator model and the pre-generated tax calculation rules, the business items currently being reported are calculated and the current reporting sample is output.
[0192] One or more embodiments of this specification provide a non-volatile computer storage medium storing computer executable instructions, which can achieve the following when executed by a computer:
[0193] Extract samples of various business items from multiple real declaration samples;
[0194] According to the calculation methods of various tax types obtained in advance, the business item samples are learned, and the calculation models of the business item samples are determined respectively, so as to realize the reuse of the business items of the real declaration samples by applying the common features in the learned calculation models to the specific business items of the actual samples;
[0195] Constructing a business item indicator model according to the calculation model of each business item sample;
[0196] According to the business item indicator model and the pre-generated tax calculation rules, the business items currently being reported are calculated and the current reporting sample is output.
[0197] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.
[0198] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0199] The above description is only one or more embodiments of this specification and is not intended to limit this specification. For those skilled in the art, one or more embodiments of this specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of this specification shall be included in the scope of the claims of this specification.
Claims
1. A method for outputting declaration samples, It is characterized in that The method comprises: Extract samples of various business items from multiple real declaration samples; According to the calculation methods of various tax types obtained in advance, the business item samples are learned, and the calculation models of the business item samples are respectively determined, so as to realize the reuse of the business items of the real declaration samples by applying the common features in the learned calculation models to the specific business items of the actual samples; Constructing a business item indicator model according to the calculation model of each business item sample; Calculate the current declared business items according to the business item indicator model and the pre-generated tax calculation rules, and output the current declared sample; The business item samples are extracted from a plurality of real declaration samples, including: Calculate the entropy value and information gain of the features in the multiple real declaration samples respectively, and match the optimal feature combination; Segmenting the multiple real declaration samples according to the optimal feature combination to form an accurate decision tree; Extracting samples of each business item according to the accurate decision tree; The business items currently being reported are calculated according to the business item indicator model and the pre-generated tax calculation rules, and the current reporting sample is output, including: Comparing the current declaration sample with the multiple real declaration samples to determine similar business items of the current declaration sample; Calculate the similar business items according to the business item indicator model and the pre-generated tax calculation rules, and output the current declaration sample; After outputting the current declaration sample, the method further includes: Analyzing difference parameters according to the comparison results of the current declared sample and the multiple real declared samples; The tax calculation rules are optimized and adjusted according to the difference parameters.
2. The method according to claim 1, It is characterized in that The entropy value and information gain of the features in the multiple real declaration samples are calculated respectively to match the optimal feature combination, including: For the features in the multiple real declaration samples, respectively calculate the entropy value of each feature, where the entropy value is used to measure the degree of confusion or uncertainty of the feature in the sample; For the features in the multiple real declaration samples, the information gain of each feature is calculated by the entropy value of each feature, and the information gain measures the degree to which the uncertainty of the result is reduced under given feature conditions; According to the information gain of each feature, the feature with the maximum information gain is selected as the optimal feature combination.
3. The method according to claim 1, It is characterized in that The step of segmenting the plurality of real declaration samples according to the optimal feature combination to form an accurate decision tree includes: Perform the first segmentation of multiple real declaration samples according to the optimal feature combination to form an initial decision tree; Cyclic segmentation of multiple real declared samples according to the optimal feature combination, and determining whether to continue segmentation according to predefined conditions, wherein the predefined conditions include a predefined maximum depth or a predefined number of samples; If the predefined conditions are met, the accurate decision tree is formed.
4. The method according to claim 1, It is characterized in that The extracting each business item sample according to the accurate decision tree includes: Inputting multiple real declaration samples into an accurate decision tree to determine the business item to which each real declaration sample belongs according to the path of each real declaration sample on the accurate decision tree; Each real declaration sample is extracted according to the business item to which it belongs, so as to extract the real declaration samples belonging to a specific business item from a plurality of real declaration sample sets to form samples of each business item.
5. The method according to claim 1, It is characterized in that The method of learning each business item sample according to the calculation method of each tax type obtained in advance, and determining the calculation model of each business item sample respectively, so as to realize the reuse of the business items of the real declaration sample by applying the common features in the learned calculation model to the specific business items of the actual sample, includes: Extracting each business item sample from the multiple real declaration samples; Based on the previously acquired tax calculation methods, the samples of each business item are studied to extract the common features of the samples; The calculation model of each business item sample is determined based on the common characteristics of the samples and the calculation methods of each tax type, so that the common characteristics in the calculation model of each business item sample can be applied to the specific business items of the actual sample to achieve the reuse of the business items of the actual declaration sample.
6. A declaration sample output device, It is characterized in that include: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to: Extract samples of various business items from multiple real declaration samples; According to the calculation methods of various tax types obtained in advance, the business item samples are learned, and the calculation models of the business item samples are respectively determined, so as to realize the reuse of the business items of the real declaration samples by applying the common features in the learned calculation models to the specific business items of the actual samples; Constructing a business item indicator model according to the calculation model of each business item sample; Calculate the current declared business items according to the business item indicator model and the pre-generated tax calculation rules, and output the current declared sample; The business item samples are extracted from a plurality of real declaration samples, including: Calculate the entropy value and information gain of the features in the multiple real declaration samples respectively, and match the optimal feature combination; Segmenting the multiple real declaration samples according to the optimal feature combination to form an accurate decision tree; Extracting samples of each business item according to the accurate decision tree; The business items currently being reported are calculated according to the business item indicator model and the pre-generated tax calculation rules, and the current reporting sample is output, including: Comparing the current declaration sample with the multiple real declaration samples to determine similar business items of the current declaration sample; Calculate the similar business items according to the business item indicator model and the pre-generated tax calculation rules, and output the current declaration sample; After outputting the current declaration sample, it also includes: Analyzing difference parameters according to the comparison results of the current declared sample and the multiple real declared samples; The tax calculation rules are optimized and adjusted according to the difference parameters.
7. A non-volatile computer storage medium, It is characterized in that Computer executable instructions are stored, and when the computer executable instructions are executed by a computer, the following can be achieved: Extract samples of various business items from multiple real declaration samples; According to the calculation methods of various tax types obtained in advance, the business item samples are learned, and the calculation models of the business item samples are respectively determined, so as to realize the reuse of the business items of the real declaration samples by applying the common features in the learned calculation models to the specific business items of the actual samples; Constructing a business item indicator model according to the calculation model of each business item sample; Calculate the current declared business items according to the business item indicator model and the pre-generated tax calculation rules, and output the current declared sample; The business item samples are extracted from a plurality of real declaration samples, including: Calculate the entropy value and information gain of the features in the multiple real declaration samples respectively, and match the optimal feature combination; Segmenting the multiple real declaration samples according to the optimal feature combination to form an accurate decision tree; Extracting samples of each business item according to the accurate decision tree; The business items currently being reported are calculated according to the business item indicator model and the pre-generated tax calculation rules, and the current reporting sample is output, including: Comparing the current declaration sample with the multiple real declaration samples to determine similar business items of the current declaration sample; Calculate the similar business items according to the business item indicator model and the pre-generated tax calculation rules, and output the current declaration sample; After outputting the current declaration sample, it also includes: Analyzing difference parameters according to the comparison results of the current declared sample and the multiple real declared samples; The tax calculation rules are optimized and adjusted according to the difference parameters.
Citation Information
Patent Citations
Taxpayer credit evaluation method based on distributed automatic feature combination
CN110827131A
Information processing method and device
CN111161080A
Decision-making method and device for data service, electronic equipment and storage medium
CN112329874A
Enterprise credit assessment automatic rule mining method and device based on improved decision tree
CN115860918A
Method and system for tax declaration
CN116051301A
