Public website text analysis method and system based on large model, and storage medium

Abstract

The invention discloses a public website page analysis processing method, a storage medium and a system, and the method comprises the following steps: S1, obtaining a public website which comprises a policy notification page, and obtaining page content data of the policy notification page; s2, analyzing the acquired page content data, and acquiring a notification file title corresponding to a page in the page content data; s3, detecting links in the policy notification page, judging whether notification file title characters and policy interpretation characters exist in a first link or not, and if yes, executing the step S4, obtaining a first page pointed by the first link, and analyzing the first page to obtain a plurality of dimension tags; by assigning the importance score of the policy, the technical effect of quickly and intuitively analyzing the importance of the policy in the page can be achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to a method for text analysis of public website pages, in particular to a method for analysis and preprocessing in response to changes in tax rate policies. BACKGROUND

[0002] Public website pages are open to the public and bear the heavy responsibility of government policy disclosure. There is a great demand for the study of public website pages. In the digital age, public websites, as an important platform for carrying public information release, policy interpretation, and public service functions, are key channels for protecting the public's right to know, participate, and supervise. Among them, the policy interpretation page is the core link to help the public accurately understand the connotation of public policy and promote the effective implementation of the policy. However, there are significant differences in the categories of current various public websites and policy interpretation pages, and there is a lack of a set of classification methods for text analysis, which poses a great obstacle to identifying fine content in the text.

[0003] Especially in terms of tax rates and related application information, such content often has strong professional and timeliness, and is often deeply bound to specific industries, business scenarios, or taxpayer types. For example, the same tax category may have multiple tax rates set according to enterprise size, revenue amount, and location, and the related application conditions may change frequently with policy adjustments. These information may be scattered in the form of textual description in the policy text, or nested in tables, formulas, or example descriptions, further increasing the difficulty of accurate identification and understanding for the public.

[0004] At the same time, existing information processing technologies have obvious limitations in dealing with such scenarios. Traditional manual sorting methods not only consume time and effort, but also are prone to information extraction errors due to individual understanding biases; some automated information extraction tools can achieve basic text analysis, but are limited by the lack of exclusive algorithm models for policy domain tax rates and application rules, making it difficult to accurately identify the logical relationship between tax rate values and application scope, exceptional cases, and other key elements, resulting in the completeness and accuracy of the extraction results cannot meet the actual application requirements, especially difficult to support fine-grained information applications in enterprise tax planning, public tax consultation, and other scenarios. SUMMARY

[0005] Therefore, it is necessary to provide a public website text analysis method based on a large model to improve the accuracy of identification. The scheme includes the following steps: S1, obtaining a public website, the public website including a policy notification page, obtaining page content data of the policy notification page,

[0006] S2, analyzing the obtained page content data to identify whether the page content data is tax-related; if so, then:

[0007] S3, extracting semantics in the text using a large model, and analyzing the scope of application, the type of tax rate, and the specific value of the tax rate in the text:

[0008] S4, obtaining existing first enterprise classification information, the first enterprise classification information including one or more of enterprise type, business description, enterprise size, applicable tax type, and applicable tax rate;

[0009] S5, extracting fields in the first enterprise classification information and combining them into text prompts, and inputting the corresponding text of the scope of application extracted in step S3 into the large model for comparison to determine whether it belongs to the scope of application;

[0010] S6, if it is determined to belong to the scope of application, then proceed to the step of updating the applicable tax type or the corresponding applicable tax rate of the first enterprise classification information according to the obtained applicable tax type, otherwise, create second enterprise classification information according to the text extracted in S3.

[0011] In some specific embodiments, the method further comprises the step of:

[0012] S7, receiving a user tax rate calculation request,

[0013] S8, obtaining user input information, the input information including enterprise operation information;

[0014] S9, calculating the applicable tax amount of the enterprise according to the enterprise operation information.

[0015] In further embodiments, after step S1, the method further comprises the step of S21, calculating the importance score of the policy notification page,

[0016] wherein the importance score step comprises: obtaining the number of all policies of the page,

[0017] calculating the importance score Dp of the pth policy: setting an initial score S for the qth responsible unit corresponding to the pth policy,

[0018]

[0019] wherein Q is the number of responsible units corresponding to the pth policy;

[0020] Nq is the total number of times the qth responsible unit appears in the responsible units of other policies;

[0021] S22, sorting each policy in descending order according to its importance score and displaying it.

[0022] A large model-based public website text analysis storage medium stores a computer program that, when executed, can perform the following steps: S1, obtaining a public website, the public website including a policy notification page, obtaining page content data of the policy notification page,

[0023] S2, analyzing the obtained page content data to identify whether the page content data is tax-related; if so, then:

[0024] S3, using a large model to extract semantics in the text and analyze the applicable scope, the type of tax rate, and the specific value of the tax rate in the text:

[0025] S4, obtaining existing first enterprise classification information, the first enterprise classification information including one or more of enterprise type, business description, enterprise size, applicable tax type, and applicable tax rate;

[0026] S5, extracting fields in the first enterprise classification information and combining them into text prompts, and inputting the text corresponding to the applicable scope in step S3 into a large model for comparison to determine whether it belongs to the applicable scope;

[0027] S6, if it is determined to belong to the applicable scope, then performing the step of updating the applicable tax type or the corresponding applicable tax rate of the first enterprise classification information based on the obtained applicable tax type, otherwise, creating second enterprise classification information based on the text extracted in S3.

[0028] In some specific embodiments, the computer program, when executed, further performs the step of,

[0029] S7, receiving a user tax rate calculation request,

[0030] S8, obtaining user input information, the input information including enterprise operation information;

[0031] S9, calculating the applicable tax amount corresponding to the enterprise based on the enterprise operation information.

[0032] In some further embodiments, after step S1, step S21 is further performed, which calculates the importance score of the policy notification page, wherein the importance score step includes: obtaining the number of all policies of the page,

[0033] calculating the importance score Dp of the pth policy: setting an initial score S for the qth responsible unit corresponding to the pth policy,

[0034]

[0035] wherein Q is the number of responsible units corresponding to the pth policy;

[0036] Nq is the total number of the qth responsibility unit appearing in other policy responsibility units;

[0037] S22, display each policy in descending order according to its importance score.

[0038] A public website text analysis system based on a large model, comprising: a website acquisition and analysis module, the website acquisition and analysis module is used for: S1, acquiring a public website, the public website includes a policy notification page, acquiring page content data of the policy notification page, S2, analyzing the acquired page content data, identifying whether the page content data is tax-related; if yes, enabling a large model extraction module, the large model extraction module is used for:

[0039] S3, extracting semantics in the text using a large model, and analyzing the applicable scope, the type of tax rate, and the specific value of the tax rate in the text:

[0040] S4, acquiring existing first enterprise classification information, the first enterprise classification information includes enterprise type, business description, enterprise size, applicable tax type, and applicable tax rate;

[0041] a tax rate application module, the tax rate application module is used for: S5, extracting the type of enterprise, the field of business description, combining into a text prompt word, and inputting the text corresponding to the applicable scope extracted in step S3 into a large model for comparison to obtain a determination of whether it belongs to the applicable scope;

[0042] S6, if it is determined to belong to the applicable scope, then proceed to: update the applicable tax type or the corresponding applicable tax rate of the first enterprise classification information according to the acquired applicable tax type, otherwise, create second enterprise classification information according to the text extracted in S3.

[0043] In some specific embodiments, the tax rate application module is further used for:

[0044] S7, receiving a user tax rate calculation request,

[0045] S8, acquiring user input information, the input information including enterprise operation information;

[0046] S9, calculating the applicable tax rate of the enterprise according to the enterprise operation information.

[0047] In some specific embodiments, further comprising an importance module, the importance module is used for when it is determined to be tax-related in step S2, performing step S21, calculating the importance score of the policy notification page,

[0048] wherein the importance score includes the steps of: acquiring the number of all policies of the page,

[0049] Calculate the importance score Dp of the pth policy: set the initial score S for the corresponding qth responsibility unit in the pth policy,

[0050]

[0051] wherein Q is the number of responsibility units corresponding to the pth policy;

[0052] Nq is the total number of times the qth responsibility unit appears in the responsibility units of other policies;

[0053] S22, display the policies in descending order of their importance scores.

[0054] Unlike the prior art, the above technical solution can help extract more fine-grained application results for more scenarios by analyzing page tax rate related text without using additional computing power using existing models. Ultimately, it can achieve better extraction results for tax rate related content in page information. BRIEF DESCRIPTION OF DRAWINGS

[0055] Figure 1 A flowchart of the public website text analysis method based on a large model described in the specific implementation;

[0056] Figure 2 A flowchart of the user tax calculation method described in the specific implementation;

[0057] Figure 3 A schematic diagram of the importance calculation step described in the specific implementation.

[0058] Figure 4 A schematic diagram of the public website text analysis storage medium based on a large model described in the specific implementation;

[0059] Figure 5 A schematic diagram of the public website text analysis system based on a large model described in the specific implementation. DETAILED DESCRIPTION

[0060] To explain the technical content, structural features, purposes and effects of the technical solutions in detail, the following will be described in detail in conjunction with the drawings.

[0061] The following detailed description is exemplary and is intended to provide further detailed description of the present application. Unless otherwise specified, all technical terms used in the present application have the same meaning as generally understood by those skilled in the art to which the present application belongs. The terms used in the present application are only for the purpose of describing the specific implementation, and are not intended to limit the exemplary embodiments according to the present application.

[0062] In the description of the present application, the phrase "and / or" is a description of a logical relationship between objects, which means that there can be three relationships, for example, A and / or B, which means that there are three cases: A exists, B exists, and A and B exist at the same time. In addition, the character " / " herein generally represents that the associated objects before and after are a "or" logical relationship.

[0063] In the present application, the terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual quantity, primary and secondary or order relationship between the entities or operations.

[0064] In the present application, without more limitation, the "include", "contain", "have" or other similar expressions used in the sentence are intended to cover non-exclusive inclusion, and the expressions do not exclude the existence of other elements in the process, method or product including the described elements, so that the process, method or product including a series of elements can not only include those limited elements, but also include other elements not explicitly listed, or also include the elements inherent to such process, method or product.

[0065] In the present application, the expressions such as "greater than", "less than", "exceed" and the like are not inclusive of the number; the expressions such as "above", "below", "within" and the like are inclusive of the number. In addition, in the description of the embodiments of the present application, the meaning of "multiple" is more than two (including two), and similar expressions related to "multiple" are also understood in this way, for example, "multiple groups", "multiple times" and the like, unless otherwise explicitly limited.

[0066] In the description of the embodiments of the present application, the spatial-related expressions used, such as "center", "longitudinal", "transverse", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "vertical", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like, indicate the orientation or position relationship based on the orientation or position relationship shown in the specific embodiment or the drawing, and are only for the convenience of describing the specific embodiments of the present application or for the reader to understand, and do not indicate or imply that the indicated device or component must have a specific position, a specific orientation, or be constructed or operated in a specific orientation, and therefore cannot be understood as a limitation on the embodiments of the present application.

[0067] Unless otherwise clearly indicated or limited by context, reference in the specification to "installation", "connection", "coupling", "fixing", "setting" or the like should be understood generically. For example, "connection" can be fixed connection, detachable connection, or integral setting; it can be mechanical connection, electrical connection, or communication connection; it can be direct connection, or indirect connection through an intermediate medium; it can be internal connection of two elements, or interaction relationship between two elements. Those skilled in the art can understand the specific meaning of the above terms in the embodiments of the present application according to the specific circumstances.

[0068] The public website refers to a website established for the purpose of providing public services, which is generally a non-profit website. In some cases of the present application, the public page website can be a website specially providing policy notification, or a departmental website with public service functions, wherein the website root domain name provides a policy notification page or a part of the link provides a policy notification page.

[0069] Please refer to Figure 1 The present application discloses a public website text analysis method based on a large model, comprising the following steps: S1, obtaining a public website, wherein the public website comprises a policy notification page, obtaining page content data of the policy notification page,

[0070] S2, analyzing the obtained page content data to identify whether the page content data is tax-related; if yes, then:

[0071] S3, extracting the semantics in the text using a large model, and analyzing the applicable scope, the type of tax rate, and the specific value of the tax rate in the text:

[0072] S4, obtaining existing first enterprise classification information, wherein the first enterprise classification information comprises one or more of enterprise type, business description, enterprise size, applicable tax type, and applicable tax rate;

[0073] S5, extracting the fields in the first enterprise classification information, combining them into a text prompt word, and inputting the text corresponding to the applicable scope extracted in step S3 into a large model for comparison to determine whether it belongs to the applicable scope;

[0074] S6, if it is determined that it belongs to the applicable scope, then proceed to the step of updating the applicable tax type or the corresponding applicable tax rate of the first enterprise classification information according to the obtained applicable tax type, otherwise, creating second enterprise classification information according to the text extracted in S3.

[0075] The large model can use chatgpt, wenxin yiyang, etc. The text can be retrieved for the scope of application, the type of tax rate, and the specific value of the tax rate. The retrieved text scope of application, tax rate type value can be a word, and the applicable tax rate is a number, etc. The existing first enterprise classification information can be a formatted enterprise information classification and tax type and tax rate correspondence, which can be pre-stored in the database, and can be input according to the user input, including enterprise type and business description field segmentation results or table input information. The classification drop-down menu option can also be manually filled in. The user's input content can also include other descriptions. Other description content allows users to input freely. In some embodiments, the user's input content can also be obtained through a dialogue guide system, and the user's voice is converted into text, and then the text is automatically recognized by a large model to obtain the enterprise type, enterprise size, turnover, and other description content. The model divides and extracts the enterprise type and business description field information. By setting the text prompt word, the accuracy of the large model in determining whether it is applicable can be improved. The model needs to consider multiple fields comprehensively to obtain the output result of whether it belongs to the scope of application. When it is determined that it can be applied, the first enterprise classification information applicable to the tax type or the corresponding applicable tax rate is updated and stored in the database. If it is found that the original first enterprise classification information in the database does not match the description in the policy, a new second enterprise classification information is generated by the large model and stored in the database, which can better meet the subsequent enterprise tax rate calculation. Through the above scheme, the policy data can be better crawled and adapted to different enterprise tax rates.

[0076] Some as Figure 2 In some further embodiments as shown in the figure, when the stored data in the database needs to be called, the following steps can be performed, S7, receiving a user tax rate calculation request,

[0077] S8, obtaining user input information, the input information including enterprise operation information;

[0078] S9, calculating the tax amount corresponding to the enterprise according to the enterprise operation information.

[0079] The user tax rate calculation request can be a user-initiated operation, such as mouse click, keyboard selection, gesture click on the display interface, etc. In specific applications, the page can jump to the input of the enterprise operation information text. In some embodiments, the user's enterprise operation information includes annual data, quarterly data, or monthly data, etc. The user can select these three items, and can also include taxable income, etc. The data to be presented in the report can be input after inputting these enterprise operation information, which can facilitate the user's calculation. In some embodiments, the scheme guides the user to input text by setting an input box, and obtains the user's input first enterprise operation information as follows:

[0080] Business type: individual industrial and commercial households;

[0081] Business description: operating community convenience stores, selling daily necessities, snacks, beverages, tobacco and alcohol, and providing express delivery services;

[0082] Enterprise size: 2 operators (no employees), annual turnover of about 300,000 yuan.

[0083] After comparing the above content with the database, the corresponding applicable tax types are obtained: value-added tax, personal income; and the applicable tax rates are calculated: value-added tax 3% (small-scale taxpayer rate), personal income tax 5%-35% (operating income over-tax progressive tax rate). The obtained results are output to the user.

[0084] In some other specific embodiments, the user's answer audio is recorded by playing preset questions through voice dialogue, the user's answer audio is recognized, and then the specific first enterprise operating information is obtained as follows:

[0085] Business type: limited liability company;

[0086] Business description: engaged in the design, production and sale of clothing, through online e-commerce platforms and offline physical stores to sell products;

[0087] Enterprise size: about 150 employees, annual turnover of about 8000 million yuan.

[0088] After comparing the above content with the third enterprise classification information in the database, the corresponding applicable tax types are obtained: value-added tax, corporate income tax, and the applicable tax rates are: value-added tax 13%, corporate income tax 25%. The above results are output to the user.

[0089] When the results within the applicable scope are obtained, the enterprise classification information and the applicable tax rate are presented to the user, or the first enterprise operating information and the applicable tax rate are combined into a matching pair and stored in the database, facilitating subsequent calling and calculation, and effectively improving the calculation efficiency of the method.

[0090] Through the above embodiments, the technical effect of calculating the tax according to the enterprise operating information input by the user is better achieved, and the demand of the user for more diversified enterprise type corresponding tax rate calculation is met.

[0091] In some other optional embodiments, please refer to Figure 3 When the page content data is obtained, steps S21 and S22 are performed simultaneously, and the importance score of the policy notification page is calculated,

[0092] The importance score step includes: obtaining the number of all policies of the page,

[0093] Calculate the importance score Dp of the pth policy: set the initial score S for the corresponding qth responsibility unit in the pth policy,

[0094]

[0095] wherein Q is the number of responsibility units corresponding to the pth policy;

[0096] Nq is the total number of times the qth responsibility unit appears in the responsibility units of other policies;

[0097] S22, display the policies in descending order of their importance scores.

[0098] At the same time of calculating the importance score, determine whether to include the tax rate adjustment related content. Yes, then proceed to step S3.

[0099] Generally speaking, in the important policy notification page, the theme is generally the file title of this notification, and at the end of the policy notification page, there is usually a link with the file title + policy interpretation. In the interpretation link, the specific analysis of the above policy can be obtained, which is usually a text analysis. Automatically obtaining the above text analysis can save the cost of manual work. After obtaining the first link, the first page pointed by the first link is text extracted, and a big data model analysis is performed to obtain dimension labels. The dimension labels can be multiple levels of dimension labels, such as title, subtitle, paragraph, body, bold, responsibility unit, policy, measure, deadline, etc. The dimension labels can be set according to the needs of the user, and the relative value of the dimension labels is the specific text extraction result. The dimension labels can have hierarchical relationships. In an embodiment, the hierarchical relationship between multiple dimension labels is marked according to the title and paragraph information, which helps to sort out the logical relationship of the text. Then analyze and sort out the responsibility unit relationship list corresponding to each policy.

[0100] In a specific embodiment, the importance of the policy needs to be evaluated, which is based on the following assumptions. When a policy has multiple responsibility units, and the fewer policies each responsibility unit is responsible for, the more specific response the policy needs, the higher the importance of the policy. The policy dimension label and the responsibility unit dimension label can have a one-to-one, one-to-many or empty relationship. S can be set to any constant according to needs. In the importance calculation, the importance score of each responsibility unit is S / Nq, and the importance score of the policy is the sum of the importance scores of all Q responsibility units corresponding to it.

[0101] In some embodiments of the present application, in order to better set the weight, the step S9 is performed: setting the initial score as Sxtq, checking whether the responsible unit q belongs to the first check list, if not, setting tq=0.5; if the check result is yes, otherwise setting tq=1.0. Adjusting the nature of the responsible unit can better assign scores to policy importance. The existence of the corresponding website of the responsible unit q and the policy notification page in the homepage position indicates the importance of the policy.

[0102] In an embodiment of the present application, the computer program further performs the following steps when executed: obtaining the corresponding website of the responsible unit q, checking whether the policy notification page exists in the homepage of the website, and setting t=1.5 if it exists.

[0103] In this way, by assigning scores to the importance of the policy, the technical effect of quickly and intuitively analyzing the importance of the policy in the page can be achieved. Then, in step S22, the policies are sorted in descending order according to their importance scores and displayed, which can provide better analysis results for the user. When the importance of the page is insufficient, a pop-up window can be sent to the user to ask whether to jump to the original policy notification page for verification.

[0104] In Figure 4 In the embodiment shown, a public website text analysis storage medium 200 based on a large model stores a computer program, which can perform the following steps when executed: S1, obtaining a public website, the public website including a policy notification page, obtaining page content data of the policy notification page,

[0105] S2, analyzing the obtained page content data to identify whether the page content data is tax-related; if yes, then:

[0106] S3, using a large model to extract the semantics in the text and analyze the applicable scope, the type of tax rate, and the specific value of the tax rate in the text:

[0107] S4, obtaining first enterprise information input by a user, the first enterprise information including enterprise type, business description, enterprise size, applicable tax type, and applicable tax rate;

[0108] S5, extracting the type of the enterprise and the field of the business description, combining them into a text prompt word, and inputting the text corresponding to the applicable scope extracted in step S3 into a large model for comparison to determine whether it belongs to the applicable scope;

[0109] S6, if it is determined that it belongs to the applicable scope, then the step of updating the applicable tax rate corresponding to the applicable tax type according to the obtained applicable tax type is performed.

[0110] Further, the computer program, when executed, further performs the step of:

[0111] S7, receiving a user tax rate calculation request,

[0112] S8, obtaining user input information, the input information including enterprise operation information;

[0113] S9, calculating the applicable tax amount corresponding to the enterprise according to the enterprise operation information.

[0114] In some further embodiments,

[0115] If it is determined in step S2 that it is tax-related, step S21 of calculating the importance score of the policy notification page is further performed, and when the importance score of the policy notification page exceeds a preset threshold, step S3 is performed.

[0116] The importance score step includes: obtaining the number of all policies of the page,

[0117] Calculating the importance score Dp of the pth policy: for the corresponding qth responsible unit in the pth policy, setting an initial score S,

[0118]

[0119] Wherein Q is the number of responsible units corresponding to the pth policy;

[0120] Nq is the total number of times the qth responsible unit appears in the responsible units of other policies;

[0121] S22, displaying each policy in descending order of its importance score.

[0122] In some embodiments of the present application, in order to better set the weight, the step of: S9, setting the initial score as Sxtq, checking whether the responsible unit q belongs to the first check list, if not, setting tq=0.5; if the check result is yes, otherwise setting tq=1.0. Adjusting the nature of the responsible unit can better assign the importance of the policy.

[0123] In an embodiment of the present application, the computer program, when executed, further performs the steps of: obtaining the corresponding website of the responsible unit q, checking whether the policy notification page exists in the homepage of the website, and if it exists, setting t=1.5.

[0124] In this way, by assigning scores to the importance scores of the policies, the technical effect of quickly and intuitively analyzing the importance of the policies in the page can be achieved. Then, in step S22, the policies are sorted in descending order according to their importance scores and displayed, which can provide better analysis results for the user. When the page importance is insufficient, a pop-up window can also be sent to the user to ask whether to jump to the original policy notification page for checking.

[0125] Please see the following Figure 5 The public website text analysis system 300 based on a large model includes a website acquisition and analysis module 31, which is configured to: S1, acquire a public website, the public website including a policy notification page, acquire page content data of the policy notification page, S2, analyze the acquired page content data, and identify whether the page content data is tax-related; if so, enable a large model extraction module 32, which is configured to:

[0126] S3, extract semantics in the text using a large model, and analyze the applicable scope, the type of tax rate, and the specific value of the tax rate in the text:

[0127] S4, acquire existing first enterprise classification information, the first enterprise classification information including enterprise type, business description, enterprise size, applicable tax type, and applicable tax rate;

[0128] A tax rate application module 33 is configured to: S5, extract the type of enterprise and the field of business description, combine them into a text prompt word, and input the text corresponding to the applicable scope extracted in step S3 into a large model for comparison to determine whether it belongs to the applicable scope;

[0129] S6, if it is determined that it belongs to the applicable scope, then proceed to: update the applicable tax type or the corresponding applicable tax rate of the first enterprise classification information according to the acquired applicable tax type, otherwise, create second enterprise classification information according to the text extracted in S3.

[0130] In some specific embodiments, the tax rate application module 33 is further configured to:

[0131] S7, receive a user tax rate calculation request,

[0132] S8, acquire user input information, the input information including enterprise operation information;

[0133] S9, calculate the applicable tax rate of the enterprise according to the enterprise operation information.

[0134] In some specific embodiments, the importance module 34 is further included for performing step S21 of calculating the importance score of the policy notification page when it is determined that the tax is relevant in step S2, and performing step S3 only when the importance score of the policy notification page exceeds a preset threshold value.

[0135] The importance score step includes: obtaining the number of all policies of the page,

[0136] The importance score Dp of the pth policy is calculated: for the qth responsible unit corresponding to the pth policy, an initial score S is set,

[0137]

[0138] Q is the number of responsible units corresponding to the pth policy;

[0139] Nq is the total number of times the qth responsible unit appears in the responsible units of other policies;

[0140] S22, the policies are sorted in descending order according to their importance scores and displayed.

[0141] In some embodiments of the present application, in order to better set the weight, the system can further perform the step of: S9, setting the initial score as Sxtq, checking whether the responsible unit q belongs to the first check list, if not, setting tq=0.5; if the check result is yes, otherwise setting tq=1.0. Adjusting the nature of the responsible unit can better assign the importance score of the policy. The existence of the corresponding website of the responsible unit q and the policy notification page in the homepage position further indicates the importance of the policy.

[0142] In an embodiment of the present application, the computer program in the system further performs the steps of obtaining the corresponding website of the responsible unit q when being run, checking whether the policy notification page exists in the homepage of the website, and setting t=1.5 if it exists.

[0143] In this way, by assigning the importance score of the policy, the technical effect of quickly and intuitively analyzing the importance of the policy in the page can be achieved. Then, step S22 of displaying the policies in descending order according to their importance scores can provide better analysis results for users. When the importance of the page is insufficient, a pop-up window can be sent to the user to ask whether to jump to the original policy notification page for checking.

[0144] It should be noted that although the above embodiments have been described herein, the patent protection scope of the utility model is not limited thereby. Therefore, based on the innovative concept of the utility model, the changes and modifications of the embodiments described herein, or the equivalent structure or equivalent process transformation made by using the contents of the utility model specification and drawings, directly or indirectly apply the above technical solutions to other related technical fields, are all included in the protection scope of the utility model patent.

Claims

1. A method for parsing text from public websites based on a large model, characterized in that, The steps include: S1, obtaining a public website, including a policy notification page, and obtaining the page content data of the policy notification page. S2. Analyze the acquired page content data to identify whether the page content data is tax-related; if so, then: S3. Extract semantics from the text using a large model, and analyze the scope of application, types of tax rates, and specific values ​​of tax rates in the text: S4. Obtain existing first enterprise classification information, which includes one or more of the following: enterprise type, business description, enterprise size, applicable tax type, and applicable tax rate. S5. Extract the fields from the first enterprise classification information, combine them into text prompts, and input them together with the text corresponding to the scope of application extracted in step S3 into the large model for comparison to obtain a determination of whether it belongs to the scope of application. S6. If it is determined that it falls within the scope of application, proceed as follows: update the applicable tax type or corresponding applicable tax rate of the first enterprise classification information according to the obtained applicable tax type; otherwise, create the second enterprise classification information according to the text extracted in S3.

2. The method for parsing public website text based on a large model according to claim 1, characterized in that, It also includes steps, S7. Receive user's tax rate calculation request. S8. Obtain user input information, including enterprise business information; S9. Calculate the applicable tax rate for the enterprise based on the enterprise's business information.

3. The method for parsing public website text based on a large model according to claim 1, characterized in that, Following step S1, step S21 is performed to calculate the importance score of the policy notification page. The importance score step includes: obtaining the total number of policies on the page. Calculate the importance score Dp of the p-th policy: For the q-th responsible unit corresponding to the p-th policy, set an initial score S. Where Q represents the number of responsible units corresponding to the p-th policy; Nq represents the total number of times the q-th responsible unit appears among the responsible units of other policies; S22. Sort the policies in descending order of their importance scores and then display them.

4. The method for parsing public website text based on a large model according to claim 3, characterized in that, The following steps are also performed: set the initial score to S×tq, verify whether the responsible unit q belongs to the first verification list, if not, set tq=0.5; if the verification result is yes, otherwise set tq=1.0; further verification steps are performed to obtain the corresponding website of the responsible unit q, and verify whether the policy notification page exists on the homepage of the website, if it exists, set t=1.

5.

5. A text parsing and storage medium for public websites based on a large model, characterized in that, The system stores a computer program that, when run, can perform the following steps: S1, obtain a public website, the public website including a policy notification page, and obtain the page content data of the policy notification page. S2. Analyze the acquired page content data to identify whether the page content data is tax-related; if so, then: S3. Extract semantics from the text using a large model, and analyze the scope of application, types of tax rates, and specific values ​​of tax rates in the text: S4. Obtain existing first enterprise classification information, which includes one or more of the following: enterprise type, business description, enterprise size, applicable tax type, and applicable tax rate. S5. Extract the fields from the first enterprise classification information, combine them into text prompts, and input them together with the text corresponding to the scope of application extracted in step S3 into the large model for comparison to obtain a determination of whether it belongs to the scope of application. S6. If it is determined that it falls within the scope of application, proceed as follows: update the applicable tax type or corresponding applicable tax rate of the first enterprise classification information according to the obtained applicable tax type; otherwise, create the second enterprise classification information according to the text extracted in S3.

6. The public website text parsing and storage medium based on a large model according to claim 5, characterized in that, The computer program also performs steps when it is run. S7. Receive user's tax rate calculation request. S8. Obtain user input information, including enterprise business information; S9. Calculate the applicable tax rate for the enterprise based on the enterprise's business information.

7. The public website text parsing and storage medium based on a large model according to claim 5, characterized in that, Following step S1, step S21 is performed to calculate the importance score of the policy notification page. The importance scoring steps include: obtaining the total number of policies on the page, Calculate the importance score Dp of the p-th policy: For the q-th responsible unit corresponding to the p-th policy, set an initial score S. Where Q represents the number of responsible units corresponding to the p-th policy; Nq represents the total number of times the q-th responsible unit appears among the responsible units of other policies; S22. Sort the policies in descending order of their importance scores and then display them.

8. A public website text parsing system based on a large model, characterized in that, include: The website acquisition and analysis module is used for: S1, acquiring public websites, including policy notification pages, and acquiring the page content data of the policy notification pages; S2, analyzing the acquired page content data to identify whether the page content data is tax-related; if so, enabling... Large model extraction module, the large model extraction module is used for: S3. Extract semantics from the text using a large model, and analyze the scope of application, types of tax rates, and specific values ​​of tax rates in the text: S4. Obtain existing first enterprise classification information, which includes enterprise type, business description, enterprise size, applicable tax type, and applicable tax rate; The tax rate application module is used for: S5, extracting the enterprise type and business description fields, combining them into text prompt words, and inputting them together with the text corresponding to the applicable scope extracted in step S3 into the large model for comparison to obtain a determination of whether it falls within the applicable scope; S6. If it is determined that it falls within the scope of application, proceed as follows: update the applicable tax type or corresponding applicable tax rate of the first enterprise classification information according to the obtained applicable tax type; otherwise, create the second enterprise classification information according to the text extracted in S3.

9. The public website text parsing system based on a large model according to claim 8, characterized in that, The tax rate application module is also used for: S7. Receive user's tax rate calculation request. S8. Obtain user input information, including enterprise business information; S9. Calculate the applicable tax rate for the enterprise based on the enterprise's business information.

10. The public website text parsing system based on a large model according to claim 8, characterized in that, It also includes an importance module, which is used to perform step S21, calculating the importance score of the policy notification page. The importance scoring steps include: obtaining the total number of policies on the page, Calculate the importance score Dp of the p-th policy: For the q-th responsible unit corresponding to the p-th policy, set an initial score S. Where Q represents the number of responsible units corresponding to the p-th policy; Nq represents the total number of times the q-th responsible unit appears among the responsible units of other policies; S22. Sort the policies in descending order of their importance scores and then display them.

Images