Order tax code determination method and device and storage medium

Abstract

The embodiment of the invention provides an order tax code determination method and device and a storage medium, and the method comprises the steps: obtaining purchase order information which comprises tax rate information, purchaser information, seller information and general ledger information; according to the tax rate information, the purchaser information, the seller information and the general ledger information, obtaining a first prediction tax code associated with the purchase order information; verifying the purchase order information and the first prediction tax code; under the condition that the verification is not passed, inputting the purchase order information into a logistic regression model, and obtaining a second prediction tax code and a confidence coefficient associated with the second prediction tax code through the logistic regression model; and when the confidence is greater than or equal to a preset confidence threshold, determining the second prediction tax code as the order tax code of the purchase order information. According to the embodiment of the invention, the automatic determination of the order tax code is realized by using an artificial intelligence means, and the efficiency of endowing the tax code to the purchase order is improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method, device and storage medium for determining an order tax code. Background of the Invention

[0003] In the field of purchase order (PO) processing, it is necessary to set the corresponding tax code according to the content of the purchase order to facilitate the company's tax management. Especially for large enterprises, various economic activities will generate different purchase orders, such as different types of orders for supply, services, etc. In terms of enterprise management, quickly assigning reasonable tax codes to purchase orders with complex contents can significantly improve the company's tax management efficiency.

[0004] In the existing corporate tax management work, a large amount of manpower is usually required to determine the order tax code. Even with the assistance of the RPA (Robotic Process Automation) system, a large amount of manual verification is required to ensure the accuracy of the tax code assigned to the purchase order. Summary of the invention

[0005] In view of this, the embodiments of the present application provide, on the one hand, a method for determining an order tax code, and on the other hand, a device for determining an order tax code and a computer-readable storage medium, so as to utilize artificial intelligence means to realize the automatic determination of the order tax code, replace the existing manual verification process, and ensure the accuracy of the tax code assigned to the purchase order while improving the efficiency of the process of assigning the tax code to the purchase order.

[0006] To solve the above technical problems, the technical solution of this application is implemented as follows:

[0007] A method for determining an order tax code comprises: obtaining purchase order information, the purchase order information comprising tax rate information, purchaser information, seller information and general ledger information; obtaining a first predicted tax code associated with the purchase order information based on the tax rate information, the purchaser information, the seller information and the general ledger information; verifying the purchase order information and the first predicted tax code; if the verification fails, inputting the purchase order information into a logistic regression model, and obtaining a second predicted tax code and a confidence level associated with the second predicted tax code through the logistic regression model; if the confidence level is greater than or equal to a preset confidence level threshold, determining the second predicted tax code as the order tax code of the purchase order information.

[0008] In one embodiment, the general ledger information includes at least one order account; obtaining a first predicted tax code associated with the purchase order information based on the tax rate information, the purchaser information, the seller information and the general ledger information includes: obtaining a mapping relationship between the tax rate information, the purchaser information, the seller information, the order account and the order tax code; and obtaining a first predicted tax code associated with the purchase order information based on the mapping relationship.

[0009] In one embodiment, the verification of the purchase order information and the first predicted tax code includes: determining whether the content of the tax rate information is the first preset information content; if the content of the tax rate information is the first preset information requirement, determining the first predicted tax code as the order tax code of the purchase order information; if the content of the tax rate information is not the first preset information content, determining whether the content of the purchaser information is the second preset information content; if the content of the purchaser information is the second preset information content, modifying the content of the first predicted tax code to the tax code content associated with the second preset information content; if the content of the purchaser information is not the second preset information content, determining the seller whether the content of the information is the third preset information content; if the content of the seller information is the third preset information content, modify the content of the first predicted tax code to the tax code content associated with the third preset information content; if the content of the seller information is not the third preset information content, determine whether the content of the general ledger information is the fourth preset information content; if the content of the general ledger information is the fourth preset information content, modify the content of the first predicted tax code to the tax code content associated with the fourth preset information content; determine the first predicted tax code after the content is modified as the order tax code of the purchase order information; if the content of the general ledger information is not the fourth preset information content, determine that the verification has failed.

[0010] In one embodiment, the method for determining the order tax code also includes: when the confidence level is less than the confidence level threshold, obtaining a revised order tax code, using the revised order tax code and the purchase order information as training samples, training the logistic regression model, and obtaining a new logistic regression model.

[0011] In one embodiment, when the general ledger information includes at least two of the order items: the first forecast tax code includes at least two first forecast sub-tax codes, and the at least two first forecast sub-tax codes correspond to at least two of the order items respectively, and the fourth preset information content includes at least two preset sub-information corresponding to at least two of the order items respectively; wherein, determining whether the content of the general ledger information is the fourth preset information content includes: determining respectively whether the content of at least two of the order items is the respective preset sub-information content; and when the content of the general ledger information is the fourth preset information content, modifying the content of the first forecast tax code to correspond to the fourth preset information The tax code content associated with the content includes: when the contents of at least two of the order items are their respective preset sub-information contents, and the tax code content associated with the preset sub-information contents of at least two of the order items are the same, the content of the first predicted tax code is modified to the tax code content associated with the preset sub-information content of any one of the order items; when the content of the general ledger information is not the fourth preset information content, determining that the verification has failed, includes: when the content of any one of the order items is not its preset sub-information content, and / or the tax code content associated with the preset sub-information content of at least two of the order items are different, determining that the verification has failed.

[0012] In one embodiment, when the general ledger information includes at least two of the order items, the second predicted tax code includes at least two second predicted sub-tax codes corresponding to the at least two order items respectively, and the confidence associated with the second predicted tax code includes at least two sub-confidences respectively associated with the at least two second predicted sub-tax codes; wherein, obtaining the second predicted tax code and the confidence associated with the second predicted tax code through the logistic regression model includes: obtaining the at least two second predicted sub-tax codes and the at least two sub-confidences through the logistic regression model; wherein, when the confidence is greater than or equal to a preset confidence threshold, determining the second predicted tax code as the order tax code of the purchase order information includes: when the contents of the at least two second predicted sub-tax codes are the same, and when the at least two sub-confidences are greater than or equal to the confidence threshold, determining any one of the at least two second predicted sub-tax codes as the order tax code of the purchase order information.

[0013] In one embodiment, the method for determining the order tax code further includes: issuing a reminder message to modify the purchase order when the contents of the at least two second predicted sub-tax codes are different and / or when any sub-confidence is less than the confidence threshold.

[0014] A device for determining an order tax code, comprising: a purchase order information acquisition module, configured to execute acquisition of purchase order information, the purchase order information including tax rate information, purchaser information, seller information and general ledger information; a first predicted tax code acquisition module, configured to execute, based on the tax rate information, the purchaser information, the seller information and the general ledger information, obtaining a first predicted tax code associated with the purchase order information; a verification module, configured to execute verification of the purchase order information and the first predicted tax code; a second predicted tax code acquisition module, configured to execute, if the verification fails, inputting the purchase order information into a logistic regression model, and obtaining a second predicted tax code and a confidence level associated with the second predicted tax code through the logistic regression model; an order tax code determination module, configured to execute, if the confidence level is greater than or equal to a preset confidence level threshold, determining the second predicted tax code as the order tax code of the purchase order information.

[0015] A device for determining an order tax code comprises at least one memory and at least one processor, wherein: the at least one memory is used to store a computer program; the at least one processor is used to call the computer program stored in the at least one memory to execute the method for determining an order tax code as described above.

[0016] A computer-readable storage medium stores a computer program thereon; the computer program can be executed by a processor and implements the method for determining an order tax code as described above.

[0017] It can be seen from the above technical solution that in the embodiment of the present application, after obtaining the purchase order information, the first predicted tax code is first obtained by using the existing mapping relationship or rules, and then the first predicted tax code is verified and modified according to the tax rate information, the purchaser information, the seller information, and the special content of the order subject to obtain the final order tax code. If the verification fails, artificial intelligence means are used to predict the second predicted tax code and the confidence associated with the second predicted tax code through a logistic regression model based on the purchase order information, and the second predicted tax code is determined as the final order tax code when the confidence is greater than or equal to the preset confidence threshold. In the embodiment of the present application, artificial intelligence means are used to deploy a robot approval role in the purchase order approval process, so as to realize the assignment of tax codes to purchase orders, which can replace manual work to correct tax codes, and realize the accuracy of the tax codes assigned to purchase orders while improving the efficiency of the process of assigning tax codes to purchase orders, and the robot's determination logic can accurately meet the preset rules. Artificial intelligence means can also continuously learn from accumulated data to enhance the model's predictive reasoning ability for tax codes.

[0018] BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to better understand the present application, the embodiments of the present application will be described in detail below with reference to the accompanying drawings, so that those skilled in the art can better understand the above and other features and advantages of the present application. In the accompanying drawings:

[0020] Figure 1 This is an exemplary flow chart of a method for determining an order tax code in an embodiment of the present application;

[0021] Figure 2 is an exemplary flow chart of obtaining a first predicted tax code according to an exemplary embodiment;

[0022] Figure 3 is an exemplary flow chart showing verification of purchase order information and a first predicted tax code according to an exemplary embodiment;

[0023] Figure 4 is an exemplary flow chart of a specific application scenario in the embodiments of the present application;

[0024] Figure 5 This is an exemplary structural diagram of a device for determining an order tax code in an embodiment of the present application;

[0025] Figure 6 This is an exemplary structural diagram of another device for determining an order tax code in an embodiment of the present application.

[0026] The reference numerals are as follows:

[0027]

[0028] Methods of implementing this application

[0029] In the tax declaration system of enterprises, due to various reasons, the rework rate in the tax declaration process is quite high. One of the most important rework activities is to change the tax code. Among them, the tax code is assigned according to the content of the purchase order. The tax code can help enterprises correctly calculate the tax of sales and purchase invoices, ensure the accuracy of financial information, and ensure the financial stability and sustainability of the enterprise. For purchase orders, tax codes can be assigned different tax rates and general ledger (GL Account) accounts. This information will be used to calculate the tax of the purchase order and ensure compliance with tax laws and tax requirements. The existing tax code determination method is to automatically assign tax codes to the content of the purchase order according to preset rules, and verify and confirm it manually. Once an error is found in the tax code assigned to the purchase order, the purchase order needs to be manually modified or rejected, which leads to a lot of manual work in the process of reworking and restarting the tax code declaration of the purchase order. In turn, these reworks also lead to the rework of other related processes, which has a significant impact on the efficiency of the enterprise's tax declaration and other work.

[0030] In view of this situation, the embodiments of the present application provide a method, device and storage medium for determining an order tax code. In view of the problem that in the above-mentioned tax reporting process, a large amount of manual rework is caused by the accuracy of tax code maintenance, which affects the tax reporting efficiency of the enterprise, a solution is proposed to assign tax codes to purchase orders by combining artificial intelligence (AI) means to replace manual verification work, thereby ensuring the accuracy of the tax codes assigned to purchase orders and improving the efficiency of the process of assigning tax codes to purchase orders.

[0031] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the technical solution of the present application is described in detail below with reference to the accompanying drawings and examples.

[0032] Figure 1 FIG. 1 is an exemplary flow chart of a method for determining an order tax code in an embodiment of the present application. Figure 1 As shown, the method for determining the order tax code mainly includes the following steps S101 to S105.

[0033] Step S101, obtaining purchase order information, the purchase order information includes tax rate information, purchaser information, seller information and general ledger information.

[0034] Step S102: Obtain a first predicted tax code associated with the purchase order information according to the tax rate information, purchaser information, seller information and general ledger information.

[0035] In an exemplary embodiment, the general ledger information includes at least one order item. Figure 2 is an exemplary flow chart of obtaining a first predicted tax code according to an exemplary embodiment. Figure 2 As shown, in an exemplary embodiment, step S102 may include the following steps S201 to S202.

[0036] Step S201, obtaining tax rate information, purchaser information, seller information, and the mapping relationship between order items and order tax codes.

[0037] Step S202: Obtain a first predicted tax code associated with the purchase order information according to the mapping relationship.

[0038] In the exemplary embodiment, the mapping relationship between the tax rate information, the purchaser information, the seller information, the order subject and the order tax code can be implemented in the form of a data table or a database. For example, it can be implemented by establishing a mapping table between the tax rate information, the purchaser information, the seller information, the order subject and the order tax code in the database. When the purchase order information is obtained, the tax code corresponding to the purchase order information, i.e., the first predicted tax code, can be obtained through the mapping table obtained from the database.

[0039] Since the contents of a purchase order usually include a wide variety of items and are complex, the first predicted tax code obtained in step S102 may not be appropriate, or the mapping relationship may not cover all situations, resulting in the inability to obtain the first predicted tax code. Therefore, after step S102, a verification process in step S103 and a process of predicting the tax code again using artificial intelligence means in step S104 are required.

[0040] Step S103: Verify the purchase order information and the first predicted tax code.

[0041] In the illustrative embodiment, step S103 is a process of verifying the reliability of the first predicted tax code obtained in step S102. The process is to determine whether the correspondence between the first predicted tax code and the tax rate information, purchaser information, seller information and general ledger information is accurate based on preset conditions. Figure 3 is an exemplary flow chart showing the verification of purchase order information and a first forecast tax code according to an exemplary embodiment. Figure 3 As shown, step S103 may specifically include the following steps S301 to S310.

[0042] Step S301: Determine whether the content of the tax rate information is the first preset information content.

[0043] Step S302: When the content of the tax rate information meets the first preset information requirement, the first predicted tax code is determined as the order tax code of the purchase order information.

[0044] In the exemplary embodiment, assuming that the first preset information content is "Z*", where "Z*" represents the content of all tax rate information starting with the letter "Z", step S302 represents that when the content of the tax rate information is the content of the tax rate information starting with the letter "Z", the first predicted tax code obtained in step S102 is the order tax code that should be finally assigned to the purchase order. Steps S301 and S302 are for the verification of the tax rate information in the purchase order.

[0045] In an illustrative embodiment, if the content of the tax rate information meets the condition of step S302, that is, the content of the tax rate information is the first preset information requirement, then after the first predicted tax code is determined as the order tax code of the purchase order information, the verification is completed and the subsequent steps are no longer executed.

[0046] Step S303: When the content of the tax rate information is not the first preset information content, determine whether the content of the purchaser information is the second preset information content.

[0047] Step S304: When the content of the purchaser information is the second preset information content, the content of the first predicted tax code is modified to the tax code content associated with the second preset information content.

[0048] Steps S303 and S304 are executed when the tax rate information verification in steps S301 and S302 fails in the purchase order, that is, when the content of the tax rate information is not the first preset information content. Taking the first preset information content as "Z*" as an example, steps S303 and S304 are executed when the content of the tax rate information is not "Z*".

[0049] In the exemplary embodiment, assuming that the second preset information content is "small-scale taxpayer", the tax code content associated with the second preset information content is "Z*", where "Z*" represents the content of all tax codes starting with the letter "Z", then step S304 represents that when the content of the purchaser information is "small-scale taxpayer", the content of the first predicted tax code is modified to "Z*" associated with "small-scale taxpayer", where the specific tax code content can be set according to actual needs. Steps S303 and S304 are for the verification of the purchaser information in the purchase order.

[0050] In an illustrative embodiment, the association between the second preset information content and the tax code content can be achieved through a mapping relationship. For example, a relationship mapping table or a database can be used to record different second preset information contents and the corresponding tax code contents in a relevant relationship mapping table or database. The tax code content corresponding to the second preset information content can be obtained by querying the relationship mapping table or the database.

[0051] Step S305: When the content of the purchaser's information is not the second preset information content, determine whether the content of the seller's information is the third preset information content.

[0052] Step S306: When the content of the seller information is the third preset information content, modify the content of the first predicted tax code to the tax code content associated with the third preset information content.

[0053] Steps S305 and S306 are executed when the verification of the purchaser information in the purchase order in steps S303 and S304 fails, that is, when the content of the purchaser information is not the second preset information content. Taking the second preset information content as "small-scale taxpayer" as an example, steps S305 and S306 are executed when the content of the purchaser information is not "small-scale taxpayer".

[0054] In the exemplary embodiment, assuming that the third preset information content is "small-scale taxpayer", the tax code content associated with the third preset information is "Z0 / J6 / P5 / Z6", where "Z0 / J6 / P5 / Z6" represents one of "Z0", "J6", "P5", and "Z6", and which one can be determined according to demand. Then step S306 represents that when the content of the seller information is "small-scale taxpayer", the content of the first predicted tax code is modified to "Z0 / J6 / P5 / Z6" associated with "small-scale taxpayer", where the specific tax code content can be set according to actual needs. Steps S305 and S306 are for the verification of the seller information in the purchase order.

[0055] In an illustrative embodiment, the association between the third preset information content and the tax code content can be achieved through a mapping relationship. For example, a relationship mapping table or a database can be used to record different third preset information contents and the corresponding tax code contents in a relevant relationship mapping table or database. The tax code content corresponding to the third preset information content can be obtained by querying the relationship mapping table or the database.

[0056] Step S307: When the content of the seller information is not the third preset information content, determine whether the content of the general ledger information is the fourth preset information content.

[0057] Step S308: When the content of the general ledger information is the fourth preset information content, modify the content of the first predicted tax code to the tax code content associated with the fourth preset information content.

[0058] Step S307 and step S308 are executed when the verification of the seller information in the purchase order in step S305 and step S306 fails, that is, when the content of the seller information is not the third preset information content.

[0059] Generally speaking, the general ledger information of a purchase order only contains one order account, which ensures that a purchase order corresponds to a unique tax code. However, due to some reasons, the general ledger information of a purchase order may contain two or even more order accounts. In the case where the general ledger information of a purchase order contains more than one order account, if the tax codes obtained based on all order accounts are the same, the purchase order can also be approved; if the tax codes obtained for each order account are different, it means that the same purchase order has different tax codes, which may mean that the same purchase order faces tax rates of different categories, making it impossible to proceed with the subsequent tax process. In this case, the purchase order needs to be rejected, and the party creating the purchase order needs to re-create different purchase orders based on different order accounts.

[0060] The following describes step S307 and step S308 by taking the simplest general ledger information containing only one order account as an example. For the case where the general ledger information contains at least two order accounts, please refer to the subsequent description.

[0061] In an illustrative embodiment, it is assumed that the fourth preset information content includes "welfare or entertainment expenses", "purchase of fixed assets", "lease and purchase of real estate", and "lease of fixed assets", and the tax code content associated with the fourth preset information includes "Z* / J0", "JK", "JL / JA", and "JO", among which "Z* / J0" represents one of "Z*" and "J0", and which one of them can be determined according to demand, and "JL / JA" represents one of "JL" and "JA", and which one of them can be determined according to demand, among which "Z* / J0" is associated with "welfare or entertainment expenses", "JK" is associated with "purchase of fixed assets", "JL / JA" is associated with "lease and purchase of real estate", and "JO" is associated with "lease of fixed assets". Then step S308 indicates that when the content of the general ledger information is "welfare or entertainment expenses", the content of the first predicted tax code is modified to "Z* / J0" associated with "welfare or entertainment expenses", where the specific tax code content can be set according to actual needs; when the content of the general ledger information is "purchase of fixed assets", the content of the first predicted tax code is modified to "JK" associated with "purchase of fixed assets"; when the content of the general ledger information is "lease and purchase of real estate", the content of the first predicted tax code is modified to "JL / JA" associated with "lease and purchase of real estate", where the specific tax code content can be set according to actual needs; when the content of the general ledger information is "lease of fixed assets", the content of the first predicted tax code is modified to "JO" associated with "lease of fixed assets". Steps S307 and S308 are verifications of the general ledger information in the purchase order.

[0062] In an illustrative embodiment, the association between the fourth preset information content and the tax code content can be achieved through a mapping relationship. For example, a relationship mapping table or a database can be used to record different fourth preset information contents and the corresponding tax code contents in a relevant relationship mapping table or database. The tax code content corresponding to the fourth preset information content can be obtained by querying the relationship mapping table or the database.

[0063] Step S309: determine the first predicted tax code after the content is modified as the order tax code of the purchase order information.

[0064] The first predicted tax code after content modification refers to the first predicted tax code after content modification in step S304, step S306 and step S308.

[0065] Step S310: When the content of the general ledger information is not the fourth preset information content, it is determined that the verification fails.

[0066] In an exemplary embodiment, when the general ledger information includes at least two order items, the first forecast tax code includes at least two first forecast sub-tax codes, and the at least two first forecast sub-tax codes correspond to the at least two order items respectively, and the fourth preset information content includes at least two preset sub-information corresponding to the at least two order items respectively. For example, when the general ledger information includes four order items of "welfare or entertainment expenses", "purchase of fixed assets", "rental and purchase of real estate", and "rental of fixed assets", the first forecast tax code includes four first forecast sub-tax codes, and the four first forecast sub-tax codes correspond to the four order items of "welfare or entertainment expenses", "purchase of fixed assets", "rental and purchase of real estate", and "rental of fixed assets", and the fourth preset information content includes four preset sub-information of "Z* / J0", "JK", "JL / JA", and "JO", wherein "Z* / J0" corresponds to "welfare or entertainment expenses", "JK" corresponds to "purchase of fixed assets", "JL / JA" corresponds to "rental and purchase of real estate", and "JO" corresponds to "rental of fixed assets".

[0067] In this case, in the exemplary embodiment, determining whether the content of the general ledger information is the fourth preset information content in step S307 includes: determining whether the content of at least two order items is the respective preset sub-information content. For example, determining whether the content of the four order items is "welfare or entertainment expenses", "purchase of fixed assets", "rental and purchase of real estate", and "rental of fixed assets" respectively.

[0068] Furthermore, in the exemplary embodiment, step S310 includes: if the content of any order item is not its preset sub-information content, and / or the tax code content associated with the preset sub-information content of at least two order items is different, determining that the verification fails. Continuing with the above example, one situation in step S310 is that if the content of any order item is not any one of "welfare or entertainment expenses", "purchase of fixed assets", "rental and purchase of real estate", "rental of fixed assets", it means that the corresponding tax code other than "welfare or entertainment expenses", "purchase of fixed assets", "rental and purchase of real estate", "rental of fixed assets" cannot be obtained, and the verification fails in this case; another situation in step S310 is that the contents of the four order items are determined to be "welfare or entertainment expenses", "purchase of fixed assets", "rental and purchase of real estate", "rental of fixed assets", but the tax code content associated with the preset sub-information content of the four order items is "Z* / J0", "JK", "JL / JA", "JO", and the contents of these four tax codes are not the same, and the verification fails in this case.

[0069] Furthermore, step S308 includes: when the contents of at least two order items are their respective preset sub-information contents, and the tax code contents associated with the preset sub-information contents of at least two order items are the same, modifying the contents of the first predicted tax code to the tax code contents associated with the preset sub-information contents of any order item. For example, the general ledger information contains two order items, "Item A" and "Item B", "Item A" and "Item B" are the preset sub-information contents of the two order items, and the tax code content associated with "Item A" is "Z1", and the tax code content associated with "Item B" is also "Z1". In this case, the content of the first predicted tax code is modified to "Z1".

[0070] Step S104: If the verification fails, the purchase order information is input into the logistic regression model, and the second predicted tax code and the confidence associated with the second predicted tax code are obtained through the logistic regression model.

[0071] In an illustrative embodiment, the logistic regression model is trained based on previous purchase orders and the corresponding correct tax codes as samples, and as new purchase orders are generated, new samples can be obtained to continuously train the logistic regression model to enhance the ability of the logistic regression model to predict tax codes for different purchase orders.

[0072] Among them, the logistic regression model is an AI algorithm model. It is a statistical model used to solve classification problems. It models the relationship between input features (such as purchase order information content) and the category to which it belongs (such as tax code) as a function, and uses this function to classify and predict new samples. The logistic regression model is based on a logistic function (such as the Sigmoid function). After linearly combining the input features, it maps the function to a probability value, which represents the probability of belonging to a certain category (such as a tax code). Logistic regression models usually use methods such as maximum likelihood estimation or gradient descent to solve model parameters.

[0073] In an illustrative embodiment, the logistic regression model may include an input layer, weights, biases, activation functions, and an output layer. The input layer contains multiple feature variables, each of which corresponds to a node. The logistic regression model uses a set of weights and biases to represent the connection strength of each node to the output node, and these connection strengths determine the degree of influence of the features on the model output. The logistic regression model can use the sigmoid function as an activation function, and the sigmoid function converts the result of the linear combination into a probability value between [0,1], representing the probability that the sample belongs to a certain category. The output layer can set multiple nodes (classifications) as needed, and the result output by each node (classification) represents the confidence (probability value) of the node (classification). In the embodiment of the present application, different classifications represent different tax codes, that is, the logistic regression model outputs the confidence of different tax codes obtained according to the purchase order. In an illustrative embodiment, in step S104, the tax code corresponding to the confidence with the highest score is determined as the second predicted tax code, and the confidence associated with the second predicted tax code is the confidence with the highest score.

[0074] Step S105: When the confidence level is greater than or equal to a preset confidence level threshold, the second predicted tax code is determined as the order tax code of the purchase order information.

[0075] In an illustrative embodiment, in order to ensure the accuracy of the order tax code obtained by the logistic regression model, the confidence threshold should be set as high as possible. In an illustrative embodiment, the confidence threshold can be set to 90%. In this way, in step S105, the second predicted tax code is determined as the order tax code of the purchase order information only when the confidence of the second predicted tax code is greater than or equal to 90%.

[0076] In an exemplary embodiment, the method for determining the order tax code of the embodiment of the present application further includes:

[0077] When the confidence level is less than the confidence level threshold, the corrected order tax code is obtained, and the corrected order tax code and purchase order information are used as training samples to train the logistic regression model to obtain a new logistic regression model.

[0078] In the illustrative embodiment, the correction of the order tax code can be achieved manually. Since the embodiment of the present application adopts the AI ​​method of the logistic regression model to determine the order tax code of the purchase order information when the verification fails, with the continuous training and improvement of the logistic regression model, it can achieve the assignment of accurate order tax codes for most purchase orders, thus greatly reducing the process of manual participation. Therefore, although there is manual correction of the order tax code, it will still improve the efficiency of determining the order tax code of the purchase order as a whole. The manual correction of the order tax code here is a useful supplement to perfecting the logistic regression model and improving the prediction effect of the logistic regression model.

[0079] In an illustrative embodiment, the process of training a logistic regression model may include: inputting a purchase order sample into the logistic regression model, obtaining a tax code classification prediction result through the logistic regression model, establishing a loss function based on the tax code classification prediction result and the tax code label of the purchase order sample, and adjusting the parameters in the logistic regression model through back propagation based on the loss function until the loss value of the loss function converges to an expected value or reaches the number of training iterations.

[0080] In an illustrative embodiment, when the general ledger information includes at least two order items, obtaining a second predicted tax code and a confidence level associated with the second predicted tax code through a logistic regression model in step S104 includes: obtaining at least two second predicted sub-tax codes and at least two sub-confidence levels through a logistic regression model.

[0081] For example, when the general ledger information includes at least two order items, the second predicted sub-tax code and the corresponding confidence level associated with each order item are obtained by performing inference on each order item through the logistic regression model.

[0082] In an illustrative embodiment, when the general ledger information includes at least two order accounts, step S105 includes: when the contents of at least two second predicted sub-tax codes are the same and when at least two sub-confidences are greater than or equal to a confidence threshold, any one of the at least two second predicted sub-tax codes is determined as the order tax code of the purchase order information.

[0083] For example, when the general ledger information contains at least two order items, if the second predicted sub-tax codes associated with each order item obtained by the logistic regression model are the same tax code, and the confidence levels of all second predicted sub-tax codes are greater than 90%, then the same tax code is determined as the order tax code for the purchase order.

[0084] In an illustrative embodiment, when the general ledger information includes at least two order items, the method for determining the order tax code of the embodiment of the present application also includes: when the contents of at least two second predicted sub-tax codes are different, and / or when any sub-confidence is less than the confidence threshold, issuing a reminder message to modify the purchase order.

[0085] For example, when the general ledger information contains at least two order items, if there are differences between the second predicted sub-tax codes associated with each order item obtained by the logistic regression model, and / or if the confidence of one second predicted sub-tax code is less than 90%, a reminder message to modify the purchase order will be issued.

[0086] In an illustrative embodiment, a reminder to modify the purchase order can be sent by sending an email to the user who submitted the purchase order. In an illustrative embodiment, if the contents of at least two second predicted sub-tax codes are different and / or if any sub-confidence is less than a confidence threshold, the purchase order can be rejected and a reminder to resubmit the purchase order can be sent.

[0087] The method for determining the order tax code of the embodiment of the present application, after obtaining the purchase order information, first uses the existing mapping relationship or rules to obtain the first predicted tax code, and then verifies and modifies the first predicted tax code according to the tax rate information, the purchaser information, the seller information, and the special content of the order account to obtain the final order tax code. If the verification fails, artificial intelligence is used to predict the second predicted tax code and the confidence associated with the second predicted tax code through a logistic regression model based on the purchase order information, and the second predicted tax code is determined as the final order tax code when the confidence is greater than or equal to a preset confidence threshold. The method for determining the order tax code of the embodiment of the present application uses artificial intelligence to assign tax codes to purchase orders, which can replace manual verification work, thereby ensuring the accuracy of the tax codes assigned to purchase orders while improving the efficiency of the process of assigning tax codes to purchase orders.

[0088] The following is a further explanation of the method for determining the order tax code of the embodiment of the present application in conjunction with a specific application scenario.

[0089] Figure 4 This is an exemplary flow chart of a specific application scenario in the embodiment of the present application. The application scenario can be a process of an economic activity within an enterprise from demand application to generation of purchase order to tax code determination and then approval feedback, such as Figure 4 As shown, the process generally includes the following steps S401 to S406.

[0090] Step S401: The requester submits a purchase order application based on the business contract.

[0091] Step S402: Verify the requester information.

[0092] Step S403: Generate a purchase order based on the contract content.

[0093] Step S404: assign a tax code to the purchase order.

[0094] Step S405: Approve and feedback to the requester.

[0095] Step S406: Execute the payment and invoicing process.

[0096] Among them, step S404 includes the following process from step S4041 to step S404B.

[0097] Step S4041, obtain purchase order information, and then proceed to step S4042.

[0098] Step S4042: Obtain the first predicted tax code according to the preset rules, and then proceed to step S4043.

[0099] The preset rule may be a mapping relationship between tax rate information, purchaser information, seller information, order account (general ledger information includes at least one order account) and order tax code in the purchase order information. One order account gives a first predicted tax code.

[0100] Step S4043: Verify the tax rate in the purchase order information and determine whether the verification is successful. If so, proceed to step S404B; otherwise, proceed to step S4044.

[0101] In the exemplary embodiment, in step S4043, it is determined whether the tax rate is "Z*". If the tax rate is "Z*", the first predicted tax code is kept unchanged and the tax rate verification is passed; if the tax rate is not "Z*", the tax rate verification is failed.

[0102] Step S4044, verify the purchaser information in the purchase order information and determine whether the verification is successful. If so, proceed to step S4049, otherwise proceed to step S4045.

[0103] In an illustrative embodiment, in step S4044, it is determined whether the content of the purchaser information is a "small-scale taxpayer". If so, the verification is successful; otherwise, the verification is unsuccessful.

[0104] Step S4045, verify the seller information in the purchase order information and determine whether the verification is successful. If so, proceed to step S4049, otherwise proceed to step S4046.

[0105] In an illustrative embodiment, in step S4045, it is determined whether the content of the seller information is a "small-scale taxpayer". If so, it indicates that the verification is passed, otherwise, it indicates that the verification is not passed.

[0106] Step S4046: Verify the general ledger information in the purchase order information and determine whether the verification is successful. If so, proceed to step S4049; otherwise, proceed to step S4047.

[0107] In an illustrative embodiment, in step S4046, it is determined whether the order account in the general ledger information is at least one of "welfare or entertainment expenses", "purchase of fixed assets", "lease and purchase of real estate", and "lease of fixed assets".

[0108] In an illustrative embodiment, in step S4046, when there is only one order account in the general ledger information, if the order account is "welfare or entertainment expenses", "purchase of fixed assets", "lease and purchase of real estate" or "lease of fixed assets", it means that the verification is passed, otherwise it means that the verification is not passed.

[0109] In an exemplary embodiment, in step S4046, when there are at least two order items in the general ledger information, if the tax codes corresponding to the at least two order items are the same, it indicates that the verification is successful, otherwise it indicates that the verification is unsuccessful.

[0110] Step S4047: input the purchase order information into the logistic regression model to obtain the second predicted tax code and the confidence associated with the second predicted tax code, and then proceed to step S4048.

[0111] Step S4048, determine whether the second predicted tax code and confidence meet the requirements, if so, proceed to step S4049, otherwise proceed to step S404A.

[0112] In an illustrative embodiment, in step S4048, if the order account in the general ledger information is one item, then the second predicted tax code obtained is one, and if the confidence of the second predicted tax code is greater than or equal to a preset confidence threshold, then the second predicted tax code and confidence meet the requirements, otherwise the second predicted tax code and confidence do not meet the requirements.

[0113] In an illustrative embodiment, in step S4048, if there are at least two order items in the general ledger information, then at least two second predicted tax codes are obtained. If the contents of at least two second predicted tax codes are the same and the confidence levels of at least two second predicted tax codes are greater than or equal to a preset confidence threshold, then the second predicted tax code and confidence level meet the requirements. Otherwise, the second predicted tax code and confidence level do not meet the requirements.

[0114] Step S4049, modify the order tax code, and then proceed to step S404B.

[0115] In the exemplary embodiment, if the purchaser information is verified in step S4044, then in step S4049, the first predicted tax code is modified to the order tax code content corresponding to the purchaser information content of "small-scale taxpayer".

[0116] In the exemplary embodiment, if the seller information is verified in step S4045, then in step S4049, the first predicted tax code is modified to the order tax code content corresponding to the seller information content of "small-scale taxpayer".

[0117] In an illustrative embodiment, if the general ledger information is verified in step S4046, then in step S4049, the first predicted tax code is modified to the order tax code content corresponding to the content of the general ledger information, wherein the modified order tax code content contains only one order tax code.

[0118] In an illustrative embodiment, if step S4048 determines that the second predicted tax code and confidence meet the requirements, then in step S4049, the content of the first predicted tax code is modified to the second predicted tax code and determined as the order tax code of the purchase order information.

[0119] Step S404A: reject the purchase order.

[0120] In an exemplary embodiment, step S404A may also include: sending an email containing reminder information of rejecting the purchase order to the user who submitted the purchase order. In an exemplary embodiment, the email may also contain reminder information requiring the user to modify the purchase order.

[0121] Step S404B: approve the purchase order.

[0122] In an exemplary embodiment, if the tax rate in the purchase order information is verified in step S4043, then in step S404B, the first predicted tax code obtained in step S4042 is determined as the order tax code of the purchase order information and the purchase order is approved.

[0123] In the exemplary embodiment, if it is the first predicted tax code obtained after modification in step S4049, then in step S404B, the first predicted tax code obtained after modification in step S404B is determined as the order tax code of the purchase order information and the purchase order is approved.

[0124] In addition, in the exemplary embodiment, in the case of rejecting a purchase order, the correct order tax code can be assigned manually according to the purchase order content and the purchase order can be approved. In this case, the approved purchase order and the manually assigned order tax code can be used as new training samples, and the logistic regression model can be further trained using the new training samples, so that the logistic regression model can assign accurate order tax codes for the new training samples and purchase orders similar to them.

[0125] The above is a detailed description of the method for determining the order tax code in the embodiment of the present invention. The following is a detailed description of the device for determining the order tax code in the embodiment of the present invention. The device for determining the order tax code in the embodiment of the present invention can be used to implement the method for determining the order tax code in the embodiment of the present invention. For details not disclosed in detail in the embodiment of the device of the present invention, please refer to the corresponding description in the embodiment of the method of the present invention.

[0126] Figure 5 FIG. 1 is an exemplary structural diagram of a device for determining an order tax code in an embodiment of the present application. Figure 5 As shown, the device may include: a purchase order information acquisition module 501, a first predicted tax code acquisition module 502, a verification module 503, a second predicted tax code acquisition module 504 and an order tax code determination module 505.

[0127] The purchase order information acquisition module 501 is configured to acquire purchase order information, which includes tax rate information, purchaser information, seller information and general ledger information.

[0128] The first predicted tax code obtaining module 502 is configured to obtain a first predicted tax code associated with the purchase order information according to the tax rate information, the purchaser information, the seller information and the general ledger information.

[0129] The verification module 503 is configured to verify the purchase order information and the first forecast tax code.

[0130] The second predicted tax code obtaining module 504 is configured to input the purchase order information into a logistic regression model when the verification fails, and obtain the second predicted tax code and the confidence associated with the second predicted tax code through the logistic regression model.

[0131] The order tax code determination module 505 is configured to determine the second predicted tax code as the order tax code of the purchase order information when the confidence level is greater than or equal to a preset confidence level threshold.

[0132] In an exemplary embodiment, the general ledger information includes at least one order account, and the first predicted tax code obtaining module 502 further includes:

[0133] A mapping relationship acquisition submodule is configured to execute acquisition of tax rate information, purchaser information, seller information, and mapping relationships between order items and order tax codes;

[0134] The first predicted tax code obtaining submodule is configured to obtain the first predicted tax code associated with the purchase order information according to the mapping relationship.

[0135] In an exemplary embodiment, the verification module 503 further includes:

[0136] A first judgment submodule is configured to judge whether the content of the tax rate information is the first preset information content;

[0137] A first determination submodule is configured to determine the first predicted tax code as the order tax code of the purchase order information when the content of the tax rate information meets the first preset information requirement;

[0138] The second judgment submodule is configured to judge whether the content of the purchaser information is the second preset information content when the content of the tax rate information is not the first preset information content;

[0139] The first modification submodule is configured to modify the content of the first predicted tax code to the tax code content associated with the second preset information content when the content of the purchaser information is the second preset information content;

[0140] A third judgment submodule is configured to judge whether the content of the seller's information is the third preset information content when the content of the purchaser's information is not the second preset information content;

[0141] The second modification submodule is configured to modify the content of the first predicted tax code to the tax code content associated with the third preset information content when the content of the seller information is the third preset information content;

[0142] A fourth judgment submodule is configured to judge whether the content of the general ledger information is the fourth preset information content when the content of the seller information is not the third preset information content;

[0143] A third modification submodule is configured to modify the content of the first predicted tax code to the tax code content associated with the fourth preset information content when the content of the general ledger information is the fourth preset information content;

[0144] The second determination submodule is configured to determine the first predicted tax code after the content modification as the order tax code of the purchase order information;

[0145] The third determining submodule is configured to determine that the verification fails when the content of the general ledger information is not the fourth preset information content.

[0146] In an exemplary embodiment, the order tax code determination device further includes:

[0147] The training module is configured to obtain the corrected order tax code when the confidence level is less than the confidence level threshold, use the corrected order tax code and purchase order information as training samples, train the logistic regression model, and obtain a new logistic regression model.

[0148] In an exemplary embodiment, when the general ledger information includes at least two order accounts: the first forecast tax code includes at least two first forecast sub-tax codes, and the at least two first forecast sub-tax codes correspond to the at least two order accounts respectively, and the fourth preset information content includes at least two preset sub-information corresponding to the at least two order accounts respectively;

[0149] The fourth judgment submodule is further configured to execute: respectively judging whether the contents of at least two order items are respective preset sub-information contents;

[0150] The third modification submodule is further configured to execute: when the contents of at least two order items are their respective preset sub-information contents, and the tax code contents associated with the preset sub-information contents of at least two order items are the same, modifying the contents of the first predicted tax code to the tax code contents associated with the preset sub-information contents of any one order item;

[0151] The third determination submodule is further configured to execute:

[0152] When the content of any order item is not its preset sub-information content, and / or the tax code contents associated with the preset sub-information contents of at least two order items are different, it is determined that the verification has failed.

[0153] In an exemplary embodiment, when the general ledger information includes at least two order items, the second predicted tax code includes at least two second predicted sub-tax codes corresponding to the at least two order items respectively, and the confidence associated with the second predicted tax code includes at least two sub-confidences associated with the at least two second predicted sub-tax codes respectively;

[0154] Wherein, the second predicted tax code obtaining module 504 is further configured to perform: obtaining at least two second predicted sub-tax codes and at least two sub-confidences through a logistic regression model;

[0155] Among them, the order tax code determination module 505 is further configured to execute: when the contents of at least two second predicted sub-tax codes are the same and at least two sub-confidences are greater than or equal to the confidence threshold, any one of the at least two second predicted sub-tax codes is determined as the order tax code of the purchase order information.

[0156] In an exemplary embodiment, the order tax code determination device further includes:

[0157] The reminder module is configured to issue a reminder message for modifying the purchase order when the contents of at least two second predicted sub-tax codes are different and / or when any sub-confidence is less than a confidence threshold.

[0158] In fact, the order tax code determination device provided by this embodiment of the present application can be implemented in various ways. For example, the order tax code determination device can be compiled into a plug-in installed in the smart terminal by using an application programming interface that complies with specific rules, or can be packaged into an application program for users to download and use.

[0159] When compiled as a plug-in, the order tax code determination device can be implemented in a variety of plug-in forms, such as ox, dll and cab. The order tax code determination device provided by this implementation of the present application can also be implemented by using specific technologies, such as Flash plug-in technology, RealPlayer plug-in technology, MMS plug-in technology, MIDI staff plug-in technology or ActiveX plug-in technology.

[0160] The method for determining the order tax code provided by this implementation of the present application can be stored in various storage media in the form of instruction storage or instruction set storage. These storage media include but are not limited to: floppy disk, optical disk, DVD, hard disk, flash memory, USB flash memory, CF card, SD card, SDHC card, MMC card, SM card, memory stick and xD card.

[0161] It should be clear that the operating system operating in the computer can implement the functions of any of the above embodiments not only by executing the program code read by the computer from the storage medium, but also by using instructions based on the program code to implement part or all of the actual operations.

[0162] For example, Figure 6 This is an exemplary structural diagram of another device for determining an order tax code in an embodiment of the present application. The device can be used to perform Figure 1 The method shown, or for implementing Figure 5 Devices in Figure 6 As shown, the device may include at least one memory 61 and at least one processor 62. In addition, some other components may also be included, such as a communication port, an input / output controller, a network communication interface, etc. These components communicate via a bus 63 or the like.

[0163] At least one memory 61 is used to store a computer program. In one example, a computer program may be understood to include Figure 6In addition, at least one memory 61 can store an operating system, etc. The operating system includes but is not limited to: Android operating system, Symbian operating system, Windows operating system, Linux operating system, etc.

[0164] At least one processor 62 is used to call the computer program stored in at least one memory 61 to execute the method for determining the order tax code described in the example of the present application. The processor 62 can be a CPU, a processing unit / module, an ASIC, a logic module or a programmable gate array, etc., and it can receive and send data through a communication port.

[0165] It should be understood that "and / or" used herein is intended to include any and all possible combinations of one or more of the associated listed items.

[0166] The number of embodiments of the present application is only used for description and does not represent the advantages of the embodiments.

[0167] It can be seen from the above technical solution that in the embodiment of the present application, after obtaining the purchase order information, the first predicted tax code is first obtained by using the existing mapping relationship or rules, and then the first predicted tax code is verified and modified according to the tax rate information, purchaser information, seller information, and special content of the order subject to obtain the final order tax code. If the verification fails, artificial intelligence means are used to predict the second predicted tax code and the confidence associated with the second predicted tax code through a logistic regression model based on the purchase order information, and the second predicted tax code is determined as the final order tax code when the confidence is greater than or equal to the preset confidence threshold. In the embodiment of the present application, artificial intelligence means are used to deploy a robot approval role in the purchase order approval process, so as to realize the assignment of tax codes to purchase orders, which can replace manual work to correct tax codes, and realize the accuracy of the tax codes assigned to purchase orders while improving the efficiency of the process of assigning tax codes to purchase orders, and the robot's determination logic can accurately meet the preset rules. Artificial intelligence means can also continuously learn from accumulated data to enhance the model's predictive reasoning ability for tax codes.

[0168] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A method for determining an order tax code, characterized in that: include: Obtain purchase order information, including tax rate information, purchaser information, seller information, and general ledger information; Obtaining a first predicted tax code associated with the purchase order information according to the tax rate information, the purchaser information, the seller information and the general ledger information; Verifying the purchase order information and the first predicted tax code; In the case where the verification fails, inputting the purchase order information into a logistic regression model, and obtaining a second predicted tax code and a confidence level associated with the second predicted tax code through the logistic regression model; When the confidence level is greater than or equal to a preset confidence level threshold, the second predicted tax code is determined as the order tax code of the purchase order information.

2. The method for determining an order tax code according to claim 1, characterized in that: The general ledger information includes at least one order account; The obtaining, according to the tax rate information, the purchaser information, the seller information and the general ledger information, a first predicted tax code associated with the purchase order information includes: Obtaining the tax rate information, the purchaser information, the seller information, the mapping relationship between the order subject and the order tax code; According to the mapping relationship, a first predicted tax code associated with the purchase order information is obtained.

3. The method for determining an order tax code according to claim 2, characterized in that: The verifying the purchase order information and the first predicted tax code includes: Determining whether the content of the tax rate information is the first preset information content; In the case where the content of the tax rate information meets the requirements of the first preset information, determining the first predicted tax code as the order tax code of the purchase order information; In the case where the content of the tax rate information is not the first preset information content, determining whether the content of the purchaser information is the second preset information content; In the case where the content of the purchaser information is the second preset information content, modifying the content of the first predicted tax code to the tax code content associated with the second preset information content; In the case that the content of the purchaser information is not the second preset information content, determining whether the content of the seller information is the third preset information content; In the case where the content of the seller information is the third preset information content, modifying the content of the first predicted tax code to the tax code content associated with the third preset information content; In a case where the content of the seller information is not the third preset information content, determining whether the content of the general ledger information is the fourth preset information content; In the case where the content of the general ledger information is the fourth preset information content, modifying the content of the first predicted tax code to the tax code content associated with the fourth preset information content; Determine the first predicted tax code after the content is modified as the order tax code of the purchase order information; When the content of the general ledger information is not the fourth preset information content, it is determined that the verification has failed.

4. The method for determining an order tax code according to claim 1, characterized in that: The method for determining the order tax code also includes: When the confidence level is less than the confidence level threshold, a revised order tax code is obtained, and the revised order tax code and the purchase order information are used as training samples to train the logistic regression model to obtain a new logistic regression model.

5. The method for determining an order tax code according to claim 3, characterized in that: When the general ledger information contains at least two of the order items: The first forecast tax code includes at least two first forecast sub-tax codes, and the at least two first forecast sub-tax codes correspond to at least two order items respectively, and the fourth preset information content includes at least two preset sub-information corresponding to at least two order items respectively; Wherein, the determining whether the content of the general ledger information is the fourth preset information content includes: respectively determining whether the contents of at least two of the order items are respective preset sub-information contents; When the content of the general ledger information is the fourth preset information content, modifying the content of the first predicted tax code to the tax code content associated with the fourth preset information content includes: In the case where the contents of at least two of the order items are their respective preset sub-information contents, and the tax code contents associated with the preset sub-information contents of at least two of the order items are the same, modifying the content of the first predicted tax code to the tax code contents associated with the preset sub-information contents of any one of the order items; When the content of the general ledger information is not the fourth preset information content, determining that the verification fails includes: When the content of any one of the order items is not its preset sub-information content, and / or the tax code contents associated with the preset sub-information contents of at least two of the order items are different, it is determined that the verification has failed.

6. The method for determining an order tax code according to claim 3, characterized in that: In the case where the general ledger information includes at least two of the order items, the second forecast tax code includes at least two second forecast sub-tax codes corresponding to the at least two order items respectively, and the confidence associated with the second forecast tax code includes at least two sub-confidences associated with the at least two second forecast sub-tax codes respectively; The obtaining of the second predicted tax code and the confidence associated with the second predicted tax code by the logistic regression model includes: Obtaining the at least two second predicted sub-tax codes and the at least two sub-confidences through the logistic regression model; Wherein, when the confidence level is greater than or equal to a preset confidence level threshold, determining the second predicted tax code as the order tax code of the purchase order information includes: When the contents of the at least two second predicted sub-tax codes are identical and when the at least two sub-confidences are both greater than or equal to the confidence threshold, any one of the at least two second predicted sub-tax codes is determined as the order tax code of the purchase order information.

7. The method for determining an order tax code according to claim 6, characterized in that: The method for determining the order tax code also includes: When the contents of the at least two second predicted sub-tax codes are different and / or when any sub-confidence is less than the confidence threshold, a reminder message for modifying the purchase order is issued.

8. A device for determining an order tax code, characterized in that: include: A purchase order information acquisition module is configured to acquire purchase order information, wherein the purchase order information includes tax rate information, purchaser information, seller information and general ledger information; A first predicted tax code obtaining module is configured to obtain a first predicted tax code associated with the purchase order information according to the tax rate information, the purchaser information, the seller information and the general ledger information; A verification module, configured to perform verification on the purchase order information and the first forecast tax code; A second predicted tax code obtaining module is configured to input the purchase order information into a logistic regression model when the verification fails, and obtain a second predicted tax code and a confidence level associated with the second predicted tax code through the logistic regression model; The order tax code determination module is configured to determine the second predicted tax code as the order tax code of the purchase order information when the confidence level is greater than or equal to a preset confidence level threshold.

9. The device for determining the order tax code is characterized in that: The invention comprises at least one memory (51) and at least one processor (52), wherein: The at least one memory (51) is used to store a computer program; The at least one processor (52) is used to call the computer program stored in the at least one memory (51) to execute the method for determining the order tax code according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon; characterized in that: The computer program can be executed by a processor and implements the method for determining an order tax code according to any one of claims 1 to 7.

Images