Data analysis method and device based on financial products

Abstract

The embodiment of the invention provides a data analysis method and device based on financial products. The method comprises the steps of obtaining basic information of a financial product to be analyzed in response to an analysis instruction of a user; wherein the analysis instruction is used for indicating to analyze the operation condition of the financial product to be analyzed, and the basic information represents the attribute of the financial product; determining target feature information according to the basic information; wherein the target feature information represents a use instruction of the financial product to be analyzed and consumption data of the financial product in a use process; obtaining analysis result information according to the target feature information; wherein the analysis result information represents the operation condition of the financial product to be analyzed. The method is used for improving the efficiency and precision of data analysis based on financial products.

Description

Technical Field

[0001] The present application relates to the financial field, and in particular to a data analysis method and device based on financial products. Background Art

[0002] Scenario finance refers to a customized financial service system built through the integration of internet technology and traditional industries within specific activity scenarios. Service providers offering related financial products need to analyze these products to generate analytical reports and better provide relevant financial services.

[0003] Existing technologies primarily rely on manual analysis to analyze data related to financial products, or on the use of general-purpose large-scale language models. However, manual analysis is often associated with high labor costs, poor timeliness, and, especially when dealing with large amounts of data, a high error rate. Using general-purpose large-scale language models also results in limited depth and accuracy in analyzing data related to financial products.

[0004] Therefore, current data analysis based on financial products has problems of low efficiency and low accuracy. Summary of the Invention

[0005] The embodiments of the present application provide a data analysis method and device based on financial products, so as to improve the efficiency and accuracy of data analysis based on financial products.

[0006] In a first aspect, embodiments of the present application provide a data analysis method based on financial products, comprising:

[0007] Responding to an analysis instruction from a user, obtaining basic information of a financial product to be analyzed; wherein the analysis instruction is used to instruct an analysis of the operating conditions of the financial product to be analyzed, and the basic information represents attributes of the financial product;

[0008] Determine target feature information based on the basic information; wherein the target feature information represents the instructions for use of the financial product to be analyzed and the consumption data of the financial product during use;

[0009] According to the target characteristic information, analysis result information is obtained; wherein the analysis result information represents the operation status of the financial product to be analyzed.

[0010] Optionally, as in the above method, determining target feature information based on basic information includes:

[0011] Determine first characteristic information and second characteristic information based on the basic information; wherein the first characteristic information represents the instructions for use of the financial product to be analyzed, and the second characteristic information represents the interaction data during use of the financial product to be analyzed;

[0012] According to the first feature information and the second feature information, determine target feature information.

[0013] Optionally, according to the above method, the first feature information is determined according to the basic information, comprising:

[0014] Based on the preset first association relationship, determine the preset document library corresponding to the basic information as the target document library; wherein, the preset first association relationship represents the association relationship between the basic information and the preset document library, and the preset document library includes a plurality of documents, and the document represents the usage instruction of the financial product to be analyzed;

[0015] For each document in the target document library, determine the paragraph at the preset position of the document as the summary information, and determine the first similarity between the summary information and the basic information; wherein, the first similarity represents the text similarity degree between the summary information and the basic information;

[0016] According to each first similarity, at least one target document is determined from the target document library;

[0017] According to the at least one target document, the first feature information is obtained based on the preset first feature extraction network; wherein, the preset first feature extraction network is used for inputting the target document and outputting the first feature information.

[0018] Optionally, according to the above method, the second feature information is determined according to the basic information, comprising:

[0019] Based on the preset second association relationship, determine the preset interface set corresponding to the basic information as the target interface set; wherein, the preset second association relationship represents the association relationship between the basic information and the preset interface set, and the preset interface set includes a plurality of identification information, and the identification information is used to represent the interface, and the interface is used to query the consumption data of the financial product corresponding to the interface in the use process;

[0020] For each identification information in the target interface set, determine the second similarity between the identification information and the basic information; wherein, the second similarity represents the text similarity degree between the identification information and the basic information;

[0021] According to each second similarity, at least one target identification information is determined from the target interface set, and the query information corresponding to the target identification information is obtained; wherein, the query information represents the consumption data of the financial product in the use process;

[0022] According to the query information corresponding to the at least one target identification information, the second feature information is obtained based on the preset second feature extraction network; wherein, the preset second feature extraction network is used for inputting the query information and outputting the second feature information.

[0023] Optionally, according to the method described above, the analysis result information is obtained according to the target feature information, comprising:

[0024] According to the target feature information, the analysis text of the financial product to be analyzed is determined;

[0025] The score information corresponding to the analysis text is determined; wherein the score information represents the standardization degree of the analysis text;

[0026] If the score information meets the preset condition, the analysis text is determined as the analysis result information of the financial product to be analyzed.

[0027] Optionally, according to the method described above, the score information corresponding to the analysis text is determined, comprising:

[0028] According to the analysis text, the first score value, the second score value, and the third score value are determined; wherein the first score value represents the case that the analysis text meets the preset rule, the second score value represents the consistency of the description content of each sentence in the analysis text, and the third score value represents the coverage of the preset keyword in the analysis text;

[0029] According to the first score value, the second score value, and the third score value, the score information is determined.

[0030] Optionally, according to the method described above, the first score value is determined according to the analysis text, comprising:

[0031] The number of preset rules is determined as the first number;

[0032] For each preset rule, according to the analysis text, the logical state information corresponding to the preset rule is determined; wherein the logical state information is the first state or the second state, the first state represents that the analysis text meets the preset rule, and the second state represents that the analysis text does not meet the preset rule;

[0033] According to the logical state information corresponding to each preset rule, the second number is determined; wherein the second number represents the number of preset rules whose logical state information is the first state;

[0034] The ratio between the first number and the second number is determined as the first score value.

[0035] Optionally, according to the method described above, the second score value is determined according to the analysis text, comprising:

[0036] According to the analysis text, based on the preset segmentation algorithm, the combination information in the analysis text is determined, and the number of combination information is determined as the third number; wherein the combination information includes field information and numerical value information corresponding to the field information, the field information represents the attribute field contained in the analysis text, and the numerical value information represents the numerical description corresponding to the attribute field;

[0037] According to the field information in each combination information, the clustering processing is performed on the combination information, and a combination set is obtained; wherein the field information of the combination information in each combination set is consistent;

[0038] For each combination set, if there is consistent numerical information, the number of the numerical information in the combination set is determined;

[0039] According to the number corresponding to each numerical information in the combination set, a fourth number is determined; wherein the fourth number is the maximum value in the number corresponding to each numerical information;

[0040] The ratio between the third number and the fourth number is determined as the second score value.

[0041] Optionally, according to the above method, the third score value is determined according to the result information, comprising:

[0042] The number of the preset keyword is determined as a fifth number;

[0043] For each preset keyword, the coverage state information corresponding to the preset keyword is determined according to the analysis text; wherein the coverage state information is the third state or the fourth state, the third state represents that the analysis text includes the preset keyword, and the fourth state represents that the analysis text does not include the preset keyword;

[0044] According to the coverage state information corresponding to each preset keyword, a sixth number is determined; wherein the sixth number represents the number of the preset keyword whose coverage state information is the third state;

[0045] The ratio between the fifth number and the sixth number is determined as the third score value.

[0046] In a second aspect, an embodiment of the present application provides a data analysis device based on a financial product, comprising:

[0047] A response unit is configured to acquire basic information of a financial product to be analyzed in response to an analysis instruction of a user; wherein the analysis instruction is used to instruct to analyze the operation situation of the financial product to be analyzed, and the basic information represents the attribute of the financial product;

[0048] A determination unit is configured to determine target feature information according to the basic information; wherein the target feature information represents the usage instruction of the financial product to be analyzed and the consumption data of the financial product in the use process;

[0049] An analysis unit is configured to obtain analysis result information according to the target feature information; wherein the analysis result information represents the operation situation of the financial product to be analyzed.

[0050] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a memory, a processor;

[0051] Memory stores computer-executable instructions;

[0052] The processor executes the computer-executable instructions stored in the memory, so that the processor executes the above first aspect and / or various possible implementations of the first aspect.

[0053] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the first aspect above and / or various possible implementation methods of the first aspect.

[0054] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the above first aspect and / or various possible implementation methods of the first aspect.

[0055] The present application provides a method and apparatus for analyzing financial product data. In response to a user's analysis instruction, the method obtains basic information about the financial product to be analyzed from the analysis instruction. Furthermore, the method determines target feature information based on the basic information, and then generates analysis result information based on the target feature information. The analysis instruction instructs the analysis of the operational status of the financial product to be analyzed. The basic information represents the attributes of the financial product, the target feature information represents the instructions for use of the financial product to be analyzed and the consumption data during use of the financial product, and the analysis result information represents the operational status of the financial product to be analyzed. It is understood that the analysis result information can be used to generate an analysis report. The present application's method reduces the high labor costs and poor timeliness associated with manual data analysis of financial products. Furthermore, the method considers combining the instructions for use of the financial product to be analyzed with the consumption data during use to analyze the financial product data, thereby addressing the issues of low depth and low accuracy of data analysis based on financial products. The present application's method is designed to improve the efficiency and accuracy of data analysis based on financial products. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0057] Figure 1 A schematic diagram of the process of a data analysis method based on financial products provided in this application Figure 1 ;

[0058] Figure 2 A schematic diagram of the process of a data analysis method based on financial products provided in this application Figure 2 ;

[0059] Figure 3 A structure diagram of a data analysis device based on a financial product Figure 1 ;

[0060] Figure 4 A structure diagram of a data analysis device based on a financial product Figure 2 ;

[0061] Figure 5 A structure diagram of an electronic device.

[0062] The specific embodiments of the present application have been shown and described in the above-described drawings, and will be described in more detail hereinafter. These drawings and written descriptions are not intended to limit the scope of the concept of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0063] The exemplary embodiments will be described in detail herein with reference to the attached drawings. In the following description, the same numbers are used to indicate the same or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not meant to represent all implementations consistent with the present application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the present application as detailed in the appended claims.

[0064] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant laws, regulations and standards, and provide corresponding operation portal for user to choose authorization or refusal.

[0065] Scenario finance refers to a customized financial service system built by integrating Internet technology with traditional industries in a specific activity scenario. For service providers who provide related financial products, it is necessary to analyze the related financial products to obtain an analysis report, and then better provide related financial services.

[0066] In the prior art, data analysis based on financial products is mainly realized by manual sorting, or using a general large language model to realize data analysis based on financial products. However, the manual sorting method has the problems of high labor cost, poor timeliness, and high error rate, especially when facing a large amount of data. The general large language model has the problems of low depth and low accuracy in data analysis based on financial products.

[0067] Therefore, the current data analysis based on financial products has problems of low efficiency and low accuracy.

[0068] The method and device for data analysis based on financial products provided by the present application obtain the basic information of the financial product to be analyzed from the analysis instruction in response to the analysis instruction of the user, further determine the target feature information according to the basic information, and then obtain the analysis result information according to the target feature information; wherein the analysis instruction is used to instruct to analyze the operation of the financial product to be analyzed, the basic information represents the attribute of the financial product, the target feature information represents the usage instruction of the financial product to be analyzed and the consumption data of the financial product in the use process, and the analysis result information represents the operation of the financial product to be analyzed. It can be understood that the analysis result information can be used to obtain an analysis report.

[0069] The method of the present application reduces the problems of high labor cost and poor timeliness in the implementation of data analysis based on financial products by manual sorting, and further considers the usage instruction of the financial product to be analyzed and the consumption data of the financial product in the use process, which are used to realize data analysis of the financial product, so as to solve the problems of low depth and low accuracy in data analysis based on financial products.

[0070] The method of the present application is used to improve the efficiency and accuracy of data analysis based on financial products.

[0071] The technical solutions of the present application and how the technical solutions of the present application solve the above technical problems will be described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes can not be described in detail in some embodiments. The embodiments of the present application will be described below with reference to the drawings.

[0072] Figure 1 Flowchart of a method for data analysis based on financial products provided by the present application Figure 1 The execution subject of the method can be a server, a host or other devices, as shown in Figure 1 The method can include:

[0073] S101, in response to the analysis instruction of the user, obtaining the basic information of the financial product to be analyzed; wherein the analysis instruction is used to instruct to analyze the operation of the financial product to be analyzed, and the basic information represents the attribute of the financial product.

[0074] Exemplarily, the analysis instruction of the user can refer to an instruction input by the user through a user interface, for example, the user selects a specific financial product in a webpage or a mobile application, or uploads input text information, the input text information can be text directly input in an input text box, or a document in a format such as TXT, PDF, Word, and the like, the input text information includes basic information of the financial product to be analyzed, and further, a “analyze” button is clicked to instruct to analyze the operation situation of the financial product to be analyzed.

[0075] The basic information of the financial product to be analyzed can represent the attributes of the financial product. Exemplarily, the basic information of the financial product to be analyzed can include, but is not limited to, attribute information such as a name, a type, an issuing institution, a product term, an interest rate, a risk level, and the like of the financial product to be analyzed.

[0076] S102, determining target feature information according to the basic information; wherein the target feature information represents the usage instruction of the financial product to be analyzed and the consumption data of the financial product in the use process.

[0077] The target feature information can represent the usage instruction of the financial product to be analyzed and the consumption data of the financial product in the use process. Exemplarily, the target feature information can be in the form of a vector.

[0078] In a possible implementation, a pre-training model is preset, the pre-training model can refer to a model pre-trained using text related to the financial field, exemplarily, the pre-training model can be a bidirectional transformer (Bert) model, a word vector (Word2Vector) model, or the like, the pre-training model is used to convert information in a text form into information in a vector form, and step S102 can include:

[0079] According to the basic information, the basic information is input into the pre-training model to obtain vector information corresponding to the output basic information, a plurality of preset feature information is acquired, wherein the preset feature information represents the usage instruction of the financial product and the consumption data of the financial product in the use process, the preset feature information is obtained by inputting the usage instruction of each financial product and the consumption data of the financial product in the use process into the pre-training model, and the similarity between the vector information corresponding to the basic information and the plurality of preset feature information is determined, if the similarity is greater than a preset feature similarity threshold (for example, 0.5), the preset feature information is determined as the target feature information.

[0080] Exemplarily, the dimension of the vector information corresponding to the basic information can be (512, 8, 206), where 512 in (512, 8, 206) represents the dimension size of the vector information corresponding to the basic information, i.e., the length of the vector information, 8 represents the number of feature vectors in the vector information corresponding to the basic information, i.e., 8 main features contained in the basic information, such as product name, product type, issuing institution, product term, interest rate, risk level, usage instructions, consumption data, etc., and 206 represents the specific content of each feature vector, i.e., each feature vector contains 206 sub-features. Exemplarily, the sub-features can be word embedding representation of text information or standardized representation of numerical information.

[0081] In an alternative embodiment, step S102 can include:

[0082] According to the basic information, determine first feature information and second feature information; wherein the first feature information represents the usage instructions of the financial product to be analyzed, and the second feature information represents the interaction data of the financial product to be analyzed in the use process; according to the first feature information and the second feature information, determine the target feature information.

[0083] The first feature information can represent the usage instructions of the financial product to be analyzed, specifically, the usage instructions of the financial product to be analyzed can include but are not limited to detailed instructions such as usage conditions, applicable scenarios, and operation processes of the financial product.

[0084] The second feature information can represent the interaction data of the financial product to be analyzed in the use process, specifically, the interaction data of the financial product to be analyzed in the use process can include but are not limited to user behavior data such as consumption amount, consumption frequency, consumption time, and consumption location of the user of the financial product.

[0085] In a possible implementation, the representation form of the first feature information and the representation form of the second feature information can both be the representation form of a vector. Exemplarily, the first feature information and the second feature information have the same dimension, for example, both are (10, 1024, 8, 206), where 10 in (10, 1024, 8, 206) represents that the first feature information is derived from 10 different data sources, 1024 represents the dimension size of the first feature information or the second feature information, i.e., the length of the first feature information or the second feature information, 8 represents the number of feature vectors in the usage instructions of the financial product to be analyzed or the interaction data of the financial product to be analyzed in the use process, i.e., 8 main features contained in the usage instructions of the financial product to be analyzed or the interaction data of the financial product to be analyzed in the use process, and 206 represents the specific content of each feature vector, i.e., each feature vector contains 206 sub-features. Exemplarily, the sub-features can be word embedding representation of text information or standardized representation of numerical information.

[0086] It can be understood that the target feature information can be obtained by fusing the first feature information and the second feature information. The fusion can refer to integrating the first feature information and the second feature information by weighted summation processing, direct splicing and the like.

[0087] Exemplarily, the dimensions of the first feature information and the second feature information are both (10, 1024, 8, 206), and the dimension of the spliced vector can be (20, 1024, 8, 206).

[0088] Exemplarily, the target feature information is determined according to the first feature information and the second feature information, and the target feature information can be represented as:

[0089] V fusion =αV struct +(1-α)V text ;

[0090] Wherein, V fusion represents the target feature information, V struct represents the first feature information, V text represents the second feature information, and a represents a preset weight factor. Exemplarily, a can also be determined by inputting the basic information into a preset convolutional neural network.

[0091] The beneficial effect of such setting is that the target feature information contains the first feature information and the second feature information, which can more comprehensively reflect the use of the financial product and the user behavior of the financial product, thereby improving the accuracy and reliability of the determination and analysis result information.

[0092] In an optional implementation, the first feature information is determined according to the basic information, which can include:

[0093] Based on the preset first association relationship, a preset document library corresponding to the basic information is determined as a target document library. The preset first association relationship represents the association relationship between the basic information and the preset document library. The preset document library includes a plurality of documents, and the document represents the use instruction of the financial product to be analyzed. For each document in the target document library, a paragraph at a preset position of the document is determined as summary information, and a first similarity between the summary information and the basic information is determined. The first similarity represents the text similarity between the summary information and the basic information. According to the first similarity, at least one target document is determined from the target document library. According to the at least one target document, the first feature information is obtained based on a preset first feature extraction network. The preset first feature extraction network is used to input the target document and output the first feature information.

[0094] The preset first association relationship can represent an association relationship between the basic information and the preset document library. In a possible implementation, the preset first association relationship can also represent an association relationship between a product name of a financial product in the basic information and the preset document library.

[0095] The preset document library can refer to a pre-set document set containing usage instructions of a financial product. For example, the documents can be official documents, user manuals, online help documents, etc. of the financial product.

[0096] In a possible implementation, the method for determining the target document library can also determine all preset document libraries as the target document library. This setting has the beneficial effect of comprehensively considering all preset document libraries, ensuring that no relevant information is missed, and improving the comprehensiveness and accuracy of data analysis.

[0097] The abstract information can refer to paragraphs in a specific position (for example, the first page of the document) of the document. It can be understood that these paragraphs contain the core content and main information of the document, and can be effectively used to determine the association degree between the document and the financial product to be analyzed.

[0098] In a possible implementation, a pre-trained model is preset. The pre-trained model can refer to a model pre-trained using texts related to the financial field. For example, the pre-trained model can be a bidirectional encoder representations from transformers (Bert) model, a word2vector model, etc. The pre-trained model is used to convert information in the form of text into information in the form of a vector. Determining the first similarity between the abstract information and the basic information can include: inputting the abstract information into the pre-trained model to obtain vector information corresponding to the output abstract information; inputting the basic information into the pre-trained model to obtain vector information corresponding to the output basic information; and determining the similarity between the vector information corresponding to the abstract information and the vector information corresponding to the basic information based on a cosine similarity algorithm, as the first similarity.

[0099] It can be understood that the cosine similarity algorithm can measure the cosine value of the angle between two vectors (vector information corresponding to the abstract information and vector information corresponding to the basic information) to measure the similarity between the two vectors.

[0100] For example, according to the first similarities, at least one target document can be determined from the target document library. This can include: determining a document with a first similarity greater than a preset document similarity threshold (for example, 0.5) as a target document; obtaining at least one target document; and if the number of target documents is less than a preset document quantity threshold (for example, 10), sorting each document in the target document library according to the first similarity from large to small, and determining the top 10 documents corresponding to the first similarity as at least one target document.

[0101] It can be understood that the number of the at least one target document is greater than or equal to the preset document quantity threshold. The beneficial effect of such setting is that both high relevance between the target document and the basic information and sufficient number of the target document are ensured, so that the first feature information related to the use instruction of the financial product can be more comprehensively extracted.

[0102] The preset first feature extraction network can refer to a neural network model for inputting the target document and outputting the first feature information. Exemplarily, the first feature extraction network can be a convolutional neural network (CNN) or a recurrent neural network (RNN) trained in the financial field. Exemplarily, the first feature extraction network can specifically be a financial field bidirectional encoder representations from transformers (FinBERT) model.

[0103] The beneficial effect of such setting is that the first feature information is derived from the target document of the use instruction of the financial product to be analyzed, and the relevance between the target document and the basic information is high, so that the first feature information can more accurately reflect the key features of the financial product, thereby improving the reliability of the analysis result information.

[0104] In an optional implementation, determining the second feature information according to the basic information can include:

[0105] determining, based on the preset second association relationship, a preset interface set corresponding to the basic information as the target interface set; wherein the preset second association relationship represents an association relationship between the basic information and the preset interface set, the preset interface set includes a plurality of identification information, the identification information is used to represent an interface, and the interface is used to query consumption data of the financial product corresponding to the interface in a use process; determining, for each identification information in the target interface set, a second similarity between the identification information and the basic information; wherein the second similarity represents a text similarity between the identification information and the basic information; determining at least one target identification information from the target interface set according to the second similarities, and obtaining query information corresponding to the target identification information; wherein the query information represents the consumption data of the financial product in the use process; obtaining the second feature information based on the preset second feature extraction network according to the query information corresponding to the at least one target identification information; wherein the preset second feature extraction network is used to input the query information and output the second feature information.

[0106] The preset second association relationship can represent an association relationship between the basic information and the preset identification information, and the identification information is used to represent an interface. Exemplarily, the identification information can be a name of the interface. In a possible implementation, the preset second association relationship can also represent an association relationship between a product name of the financial product in the basic information and the preset identification information.

[0107] The preset interface set can refer to a set of interfaces that are pre-configured to query consumption data of the financial product in use. Exemplarily, the interfaces correspond to structured data as query information, and the query information represents the consumption data of the financial product in use. Exemplarily, the query information includes, but is not limited to, transaction serial numbers, amounts, timestamps, transaction parties, and the like.

[0108] In a possible implementation, the method for determining the target interface set can also determine all the preset interface sets as the target document library. This setting has the beneficial effect of comprehensively considering all the preset interface sets, ensuring that no relevant information is missed, and improving the comprehensiveness and accuracy of data analysis.

[0109] In a possible implementation, the pre-training model is preset, and the pre-training model can refer to a model that is pre-trained using text related to the financial field. Exemplarily, the pre-training model can be a bidirectional encoder representations from transformers (Bert) model, a Word2Vector model, or the like. The pre-training model is used to convert information in the form of text into information in the form of a vector. Determining the second similarity between the identification information and the basic information can include: inputting the identification information into the pre-training model to obtain vector information corresponding to the output identification information; inputting the basic information into the pre-training model to obtain vector information corresponding to the output basic information; and determining, based on a cosine similarity algorithm, a similarity between the vector information corresponding to the identification information and the vector information corresponding to the basic information as the second similarity.

[0110] It can be understood that the cosine similarity algorithm can measure the cosine value of the angle between two vectors (vector information corresponding to the abstract information and vector information corresponding to the basic information) to measure the similarity between the two vectors.

[0111] Exemplarily, determining, according to the second similarities, at least one target identification information from the target interface set can include: determining identification information with a second similarity greater than a preset interface similarity threshold (for example, 0.5) as target identification information; obtaining at least one target identification information; and if the number of the at least one target identification information is less than a preset interface quantity threshold (for example, 10), sorting each document in the target interface set according to the second similarity from large to small, and determining identification information corresponding to the top 10 second similarities as the at least one target identification information.

[0112] Further, after the at least one target identification information is determined, the query information corresponding to the target identification information can be obtained. It can be understood that the number of the at least one target identification information is greater than or equal to the preset interface quantity threshold. The beneficial effect of such setting is that it not only ensures the high correlation between the target identification information and the basic information, but also ensures that the number of the target identification information is sufficient, that is, the number of the query information corresponding to the target identification information is sufficient, so that the second feature information related to the consumption data of the financial product in the use process can be more comprehensively extracted.

[0113] The preset second feature extraction network can refer to a neural network model for inputting query information and outputting second feature information. Exemplarily, the second feature extraction network can be a convolutional neural network (CNN) or a recurrent neural network (RNN) trained in the financial field. Exemplarily, the second feature extraction network can be a financial field bidirectional encoder (FinBERT) model.

[0114] In a possible implementation, before the second feature information is obtained based on the preset second feature extraction network according to the query information corresponding to the at least one target identification information, the method can further include: sequentially performing cleaning processing and structural processing on the query information corresponding to the at least one target identification information to obtain updated query information corresponding to the at least one target identification information; and obtaining the second feature information based on the preset second feature extraction network according to the updated query information corresponding to the at least one target identification information.

[0115] The cleaning processing can refer to removing noise data, duplicate data, missing data, etc. in the query information to improve the quality of the query information. Exemplarily, the cleaning processing can include but is not limited to removing query information with an empty transaction serial number, removing query information with a negative amount, etc.

[0116] The structural processing can refer to converting the query information into a unified format so that the accuracy of the obtained second feature information is higher. Exemplarily, the amount can be converted into a unified currency unit (for example, yuan of RMB), the timestamp can be converted into a unified date and time format (for example, “YYYY-MM-DD HH:MM:SS”), etc.

[0117] The beneficial effect of such setting is that the second feature information is derived from the query information corresponding to the target identification information, and the query information represents the consumption data of the financial product in the use process, the correlation between the target identification information and the basic information is high, so that the second feature information can more accurately reflect the key features of the financial product, thereby improving the reliability of the obtained analysis result information.

[0118] S103, obtaining analysis result information according to the target feature information; wherein the analysis result information represents the operation situation of the financial product to be analyzed.

[0119] The analysis result information can refer to comprehensive data for evaluating and describing the performance of the financial product to be analyzed in the market, user acceptance, risk level, transaction activity, and other multi-dimensional operation indicators. The analysis result information can be in the form of pure text, or in the form of a chart, a report, or other forms. It can be understood that the analysis result information can help the service provider of the financial product to understand the overall operation situation of the financial product, so as to make corresponding decisions.

[0120] For example, a reasoning large model (Reasoning LLMs) is preset, which is used to generate analysis result information according to target feature information; a reasoning knowledge base and a reasoning prompt word are preset to assist the reasoning large model in accurately analyzing the target feature information. The reasoning knowledge base can refer to a database containing professional knowledge in the financial field, historical data, market trends, and the like. The reasoning prompt word can refer to a keyword or instruction used to guide the reasoning large model to perform specific analysis, such as "user satisfaction analysis", "risk assessment", "market performance scoring", and the like.

[0121] According to the target feature information, the analysis result information is obtained, which can include inputting the target feature information into the preset reasoning large model to obtain the output analysis result information. Specifically, the reasoning large model outputs the analysis result information according to the target feature information, the preset reasoning knowledge base, and the reasoning prompt word.

[0122] The method of the present application reduces the high labor cost and poor timeliness of manual sorting in the implementation of data analysis based on financial products. In addition, the method also considers combining the usage instructions of the financial product to be analyzed and the consumption data of the financial product in the use process to realize data analysis of the financial product, so as to solve the problem of low depth and low accuracy of data analysis based on financial products. The method of the present application is used to improve the efficiency and accuracy of data analysis based on financial products.

[0123] Figure 2 A flowchart of a data analysis method based on a financial product provided by the present application Figure 2 The execution subject of the method can be a server, a host, or other devices, as shown in Figure 2 The method can include:

[0124] S201, in response to an analysis instruction of a user, obtaining basic information of a financial product to be analyzed; wherein the analysis instruction is used to instruct to analyze the operation situation of the financial product to be analyzed, and the basic information represents the attributes of the financial product.

[0125] Exemplarily, the step can be referred to the step S101 described above, and will not be repeated here.

[0126] S202, determining target feature information according to the basic information; wherein the target feature information represents the usage instructions of the financial product to be analyzed and the consumption data of the financial product in the use process.

[0127] Exemplarily, the step can be referred to the step S102 described above, and will not be repeated here.

[0128] S203, determining the analysis text of the financial product to be analyzed according to the target feature information.

[0129] The analysis text can refer to comprehensive data for evaluating and describing the multi-dimensional operation indicators of the financial product to be analyzed, such as performance in the market, user acceptance, risk level, transaction activity, etc. The performance form of the analysis text is pure text form.

[0130] Exemplarily, there are pre-set reasoning large models (Reasoning LLMs) for generating the analysis text of the financial product to be analyzed according to the target feature information; there are pre-set reasoning knowledge bases and reasoning prompt words for assisting the reasoning large models to accurately analyze the target feature information, wherein the reasoning knowledge base can refer to a database containing professional knowledge in the financial field, historical data, market trends, etc. The reasoning prompt word can refer to a keyword or instruction for guiding the reasoning large model to perform specific analysis, such as “user satisfaction analysis”, “risk assessment”, “market performance scoring”, etc.

[0131] According to the target feature information, the analysis result information can include: inputting the target feature information into the pre-set reasoning large model to obtain the output analysis text of the financial product to be analyzed. Specifically, the reasoning large model outputs the analysis text of the financial product to be analyzed according to the target feature information, the pre-set reasoning knowledge base and the reasoning prompt word.

[0132] S204, determining score information corresponding to the analysis text; wherein the score information represents the standardization degree of the analysis text.

[0133] The score information can represent the standardization degree of the analysis text, and it can also be understood that the score information can represent the quality of the analysis text.

[0134] In a possible implementation, a quality detection module is preset, and the quality detection module includes three sub-modules, namely a financial logic verification layer, a data consistency verification layer, and a scenario relevance analysis layer. The financial logic verification layer is configured to verify preset rules (for example, cash flow logic, risk control index compliance, and the like). It can be understood that the more the number of the analysis text meeting the preset rules, the higher the numerical value of the score output by the financial logic verification layer. The data consistency verification layer is configured to verify whether the context in the analysis text is consistent. It can be understood that the higher the text consistency of the context in the analysis text, the higher the numerical value of the score output by the data consistency verification layer. The scenario relevance analysis layer is configured to verify the coverage of preset keywords (for example, scenario-specific financial terms). It can be understood that the higher the coverage of the preset keywords, the higher the numerical value of the score output by the scenario relevance analysis layer. Further, the quality detection module can be configured to perform weighted summation processing on the scores output by the financial logic verification layer, the data consistency verification layer, and the scenario relevance analysis layer, and obtain score information corresponding to the analysis text.

[0135] In an optional implementation, step S204 can include:

[0136] According to the analysis text, a first score value, a second score value, and a third score value are determined. The first score value represents a case where the analysis text meets preset rules, the second score value represents a consistent case of description content of each sentence in the analysis text, and the third score value represents a coverage case of preset keywords in the analysis text. According to the first score value, the second score value, and the third score value, score information is determined.

[0137] The first score value can represent a case where the analysis text meets preset rules, the second score value can represent a consistent case of description content of each sentence in the analysis text, and the third score value can represent a coverage case of preset keywords in the analysis text.

[0138] Further, the first score value, the second score value, and the third score value can be subjected to weighted summation processing to obtain score information. The weighted summation processing can refer to multiplying each score value by a corresponding weight according to a preset weight corresponding to each of the three score values, adding the score values after multiplying the weights, and obtaining score information.

[0139] The beneficial effect of such a setting is that, through the weighted summation processing, the score values of different dimensions can be comprehensively considered, and it is ensured that the final score information can fully reflect the quality of the analysis text.

[0140] In an optional implementation, according to the analysis text, a first score value can be determined, which can include:

[0141] determining a first number of the preset rules; for each of the preset rules, determining, according to the analysis text, logical state information corresponding to the preset rule; wherein the logical state information is a first state or a second state, the first state indicating that the analysis text meets the preset rule, and the second state indicating that the analysis text does not meet the preset rule; determining a second number according to the logical state information corresponding to each of the preset rules; wherein the second number represents the number of the preset rules whose logical state information is the first state; and determining a first score value as a ratio between the first number and the second number.

[0142] The preset rule can refer to a rule predefined by a staff according to a financial logic, a compliance requirement, a risk control, etc., and used for evaluating a quality of the analysis text. The preset rule can include, but is not limited to, that a calculation of an amount in the analysis text meets a preset calculation formula, that the analysis text contains a relevant compliance legal provision, and that the analysis text contains a preset risk control index-related keyword.

[0143] It can be understood that for each of the preset rules, the logical state information corresponding to the preset rule can be determined according to the analysis text. The logical state information is the first state (indicating that the analysis text meets the preset rule) or the second state (indicating that the analysis text does not meet the preset rule). By determining the number of the analysis text meeting the preset rule and the number of the analysis text not meeting the preset rule, the first score value can be determined.

[0144] Exemplarily, the first number is the number of the preset rules, and the second number is the number of the preset rules whose logical state information is the second state. The first score value can be obtained by dividing the second number by the first number.

[0145] The beneficial effect of such a setting is that by judging the logical state information of the preset rule and then determining the first score value, the quality of the analysis text can be effectively evaluated, and the depth and precision of the data analysis based on the financial product can be improved.

[0146] In an optional implementation, the second score value can be determined according to the analysis text, which can include:

[0147] According to the analysis text, based on the preset segmentation algorithm, determine the combination information in the analysis text, and determine the number of combination information, as the third number; wherein, the combination information includes field information and numerical information corresponding to the field information, the field information represents the attribute field contained in the analysis text, and the numerical information represents the numerical description corresponding to the attribute field; according to the field information in each combination information, the clustering processing is carried out on each combination information, and the combination set is obtained; wherein, the field information of the combination information in each combination set is consistent; for each combination set, if there is consistent numerical information, the number of numerical information in the combination set is determined; according to the number corresponding to each numerical information in the combination set, the fourth number is determined; wherein, the fourth number is the maximum value in the number corresponding to each numerical information; the ratio between the third number and the fourth number is determined as the second score value.

[0148] The preset segmentation algorithm can include but is not limited to regular expression, natural language processing tool NLTK (Natural Language Toolkit) and the like.

[0149] The combination information can include field information and numerical information corresponding to the field information, the field information represents the attribute field contained in the analysis text, and the numerical information represents the numerical description corresponding to the attribute field. Exemplarily, one combination information is [enterprise number, 100], wherein, the enterprise number is the field information, and 100 is the numerical information.

[0150] In a possible implementation, the field information is included in the preset field text library. Exemplarily, the preset field text library can be a database containing all possible field information pre-arranged by the staff, which is used to ensure the accuracy and consistency of the extracted field information.

[0151] In a possible implementation, the combination information includes at least one field information and numerical information corresponding to the field information. Exemplarily, the at least one field information and numerical information corresponding to the field information can be represented as {[enterprise number, 100], [user maximum age, 70], [user male proportion, 0.4]}.

[0152] Further, according to the field information in each combination information, the clustering processing can be carried out on each combination information, and the combination set is obtained; wherein, the field information of the combination information in each combination set is consistent.

[0153] The clustering processing can include but is not limited to k-means algorithm, density-based clustering algorithm (DBSCAN, Density-Based Spatial Clustering of Applications with Noise) and the like.

[0154] It can be understood that in a combination set, the field information of each combination information in the combination set is consistent (for example, all are "enterprise number"), and the numerical information of each combination information in the combination set is not necessarily consistent.

[0155] Exemplarily, the combination information in a combination set can be [enterprise number, 100], [enterprise number, 100], [enterprise number, 100], [enterprise number, 1000] in turn, wherein the number of the numerical information 100 in the combination set is 3 (i.e. the number corresponding to the numerical information 100), the number of the numerical information 1000 in the combination set is 1 (i.e. the number corresponding to the numerical information 1000), and the maximum value of the number corresponding to each numerical information is 3. Taking 3 as the fourth number and the number 4 of the combination information as the third number, the second score value can be obtained by dividing the fourth number 3 by the third number 4.

[0156] The beneficial effect of such setting is that by determining the consistency of the combination information in the analysis text and then determining the second score value, the quality of the analysis text can be effectively evaluated, and the depth and precision of the data analysis based on the financial product can be improved.

[0157] In an optional implementation, according to the result information, determining the third score value can include:

[0158] determining the number of the preset keywords as a fifth number; for each preset keyword, determining the coverage state information corresponding to the preset keyword according to the analysis text; wherein the coverage state information is the third state or the fourth state, the third state represents that the analysis text includes the preset keyword, and the fourth state represents that the analysis text does not include the preset keyword; determining a sixth number according to the coverage state information corresponding to each preset keyword; wherein the sixth number represents the number of the preset keywords whose coverage state information is the third state; and determining the ratio between the fifth number and the sixth number as the third score value.

[0159] The preset keywords can include but are not limited to scene financial related proper nouns. The preset keywords are set by the staff in advance.

[0160] It can be understood that for each preset keyword, the coverage state information corresponding to the preset keyword can be determined according to the analysis text, and the coverage state information is the third state (the analysis text includes the preset keyword) or the fourth state (representing that the analysis text does not include the preset keyword). By determining the number of the analysis text including the preset keyword and the number of the analysis text not including the preset keyword, the third score value can be determined.

[0161] Exemplarily, the fifth quantity is a quantity of preset keywords, the sixth quantity is a quantity of the preset keywords whose coverage state information is the fourth state, and the third score value can be obtained by dividing the sixth quantity by the fifth quantity.

[0162] The beneficial effect of such an arrangement is that by judging the coverage state information of the preset keywords and then determining the third score value, the quality of the analysis text can be effectively evaluated and analyzed, thereby improving the depth and accuracy of data analysis based on the financial product.

[0163] S205, if the score information meets the preset condition, the analysis text is determined as the analysis result information of the financial product to be analyzed.

[0164] Exemplarily, the preset condition can mean that the score information reaches or exceeds a preset score threshold (for example, 0.85), and the preset condition is used to judge whether the quality of the analysis text meets the requirements. The higher the score information, the higher the quality of the analysis text, and the higher the depth and accuracy of data analysis based on the financial product.

[0165] Exemplarily, if the score information is greater than or equal to the preset score threshold, the analysis text is determined as the analysis result information of the financial product to be analyzed.

[0166] In one possible implementation method, if the score information does not meet the preset condition, the step of determining the target feature information according to the basic information is re-executed until the score information meets the preset condition.

[0167] In one possible implementation method, a general model (for example, a large language model) is preset, and a generated knowledge base, a report generation template, and a report generation prompt word are also preset. The generated knowledge base can refer to a database containing information such as professional knowledge in the financial field, historical data, market trends, and industry standards, which is used to provide background knowledge and reference information to generate more comprehensive and accurate analysis reports. The report generation template can refer to a predefined report structure and format, including but not limited to cover, abstract, introduction, analysis result, conclusion, chart, and report, to ensure that the generated report is consistent and professional. The report generation prompt word can refer to keywords or instructions that guide the general model to generate specific content.

[0168] After determining the analysis result information of the financial product to be analyzed, the analysis result information of the financial product to be analyzed, the preset generated knowledge base, the report generation template, and the report generation prompt word can be input into the preset general model to obtain an analysis report of the financial product to be analyzed.

[0169] The method of the present application reduces the high labor cost and poor timeliness of manual carding in the implementation of financial product-based data analysis, and also considers combining the use instructions of the financial product to be analyzed and the consumption data of the financial product in the use process to realize data analysis of the financial product, so as to solve the problems of low depth and low accuracy in financial product-based data analysis. At the same time, score information is introduced, and if the score information meets the preset condition, the analysis text is determined as the analysis result information of the financial product to be analyzed, so as to ensure the high quality and reliability of the analysis result information. The method of the present application is used to improve the efficiency and accuracy of financial product-based data analysis.

[0170] Figure 3 A structure diagram of a financial product-based data analysis device provided by the present application Figure 1 As shown in Figure 3 , the financial product-based data analysis device 30 comprises a response unit 301, a determination unit 302, and an analysis unit 303.

[0171] The response unit 301 is configured to acquire basic information of a financial product to be analyzed in response to an analysis instruction of a user; wherein the analysis instruction is used to instruct to analyze the operation situation of the financial product to be analyzed, and the basic information represents the attributes of the financial product;

[0172] The determination unit 302 is configured to determine target feature information according to the basic information; wherein the target feature information represents the use instructions of the financial product to be analyzed and the consumption data of the financial product in the use process;

[0173] The analysis unit 303 is configured to obtain analysis result information according to the target feature information; wherein the analysis result information represents the operation situation of the financial product to be analyzed.

[0174] Figure 4 A structure diagram of a financial product-based data analysis device provided by the present application Figure 2 As shown in Figure 4 , the financial product-based data analysis device 40 comprises a response unit 401, a determination unit 402, and an analysis unit 403, wherein the determination unit 402 further comprises a first processing module 4021 and a second processing module 4022, and the analysis unit 403 further comprises a third processing module 4031, a fourth processing module 4032, and a fifth processing module 4033.

[0175] The first processing module 4021 is configured to determine first feature information and second feature information according to the basic information; wherein the first feature information represents the use instructions of the financial product to be analyzed, and the second feature information represents the interaction data of the financial product to be analyzed in the use process;

[0176] The second processing module 4022 is configured to determine the target feature information according to the first feature information and the second feature information.

[0177] In an optional example, the first processing module 4021 includes a first sub-module and a second sub-module.

[0178] The first sub-module is configured to determine a preset document library corresponding to the basic information as a target document library based on a preset first association relationship, wherein the preset first association relationship represents an association relationship between the basic information and the preset document library, the preset document library includes a plurality of documents, and each document represents a use instruction of a financial product to be analyzed; for each document in the target document library, a paragraph at a preset position of the document is determined as summary information, and a first similarity between the summary information and the basic information is determined, wherein the first similarity represents a text similarity between the summary information and the basic information; at least one target document is determined from the target document library according to the first similarities; and the first feature information is obtained based on a preset first feature extraction network according to the at least one target document, wherein the preset first feature extraction network is configured to input the target document and output the first feature information.

[0179] The first sub-module is configured to determine a preset interface set corresponding to the basic information as a target interface set based on a preset second association relationship, wherein the preset second association relationship represents an association relationship between the basic information and the preset interface set, the preset interface set includes a plurality of identification information, and each identification information represents an interface, and the interface is used to query consumption data of a financial product corresponding to the interface in a use process; for each identification information in the target interface set, a second similarity between the identification information and the basic information is determined, wherein the second similarity represents a text similarity between the identification information and the basic information; at least one target identification information is determined from the target interface set according to the second similarities, and query information corresponding to the target identification information is obtained, wherein the query information represents the consumption data of the financial product in the use process; and the second feature information is obtained based on a preset second feature extraction network according to the query information corresponding to the at least one target identification information, wherein the preset second feature extraction network is configured to input the query information and output the second feature information.

[0180] The third processing module 4031 is configured to determine an analysis text of the financial product to be analyzed according to the target feature information.

[0181] The fourth processing module 4032 is configured to determine score information corresponding to the analysis text, wherein the score information represents a standard degree of the analysis text.

[0182] The fifth processing module 4033 is configured to determine the analysis text as analysis result information of the financial product to be analyzed if the score information satisfies a preset condition.

[0183] In an optional example, the fourth processing module 4032 is further configured to determine, according to the analysis text, a first score value, a second score value, and a third score value, wherein the first score value represents a condition that the analysis text meets the preset rule, the second score value represents a consistency condition of description content of each sentence in the analysis text, and the third score value represents a coverage condition of the preset keyword in the analysis text; and determine the score information according to the first score value, the second score value, and the third score value.

[0184] In an optional example, the fourth processing module 4032 is further configured to determine a number of the preset rules as a first number, determine, for each preset rule, logical state information corresponding to the preset rule according to the analysis text, wherein the logical state information is a first state or a second state, the first state represents that the analysis text meets the preset rule, and the second state represents that the analysis text does not meet the preset rule, determine a second number according to the logical state information corresponding to each preset rule, wherein the second number represents a number of the preset rules with the first state, and determine a ratio between the first number and the second number as the first score value.

[0185] In an optional example, the fourth processing module 4032 is further configured to determine, according to the analysis text, combined information in the analysis text based on a preset segmentation algorithm, and determine a number of the combined information as a third number, wherein the combined information includes field information and numerical value information corresponding to the field information, the field information represents an attribute field contained in the analysis text, and the numerical value information represents a numerical description corresponding to the attribute field, perform clustering processing on each combined information according to the field information in the combined information to obtain a combined set, wherein the field information of the combined information in each combined set is consistent, for each combined set, determine a number of the numerical value information in the combined set if there is consistent numerical value information, determine a fourth number according to the number of each numerical value information in the combined set, wherein the fourth number is a maximum value in the number of each numerical value information, and determine a ratio between the third number and the fourth number as the second score value.

[0186] In an optional example, the fourth processing module 4032 is further configured to determine a number of the preset keywords as a fifth number, determine, for each preset keyword, coverage state information corresponding to the preset keyword according to the analysis text, wherein the coverage state information is a third state or a fourth state, the third state represents that the analysis text includes the preset keyword, and the fourth state represents that the analysis text does not include the preset keyword, determine a sixth number according to the coverage state information corresponding to each preset keyword, wherein the sixth number represents a number of the preset keywords with the third state, and determine a ratio between the fifth number and the sixth number as the third score value.

[0187] Figure 5 A structural schematic diagram of an electronic device provided in the present application is shown in FIG. 1. As shown in the figure, the electronic device 50 provided in the present embodiment comprises at least one processor 501 and a memory 502. Optionally, the device 50 further comprises a communication component 503. The processor 501, the memory 502 and the communication component 503 are connected through a bus 504. Figure 5

[0188] In the implementation process, the at least one processor 501 executes the computer-executable instructions stored in the memory 502, so that the at least one processor 501 executes the above-mentioned method.

[0189] The specific implementation process of the processor 501 can refer to the above-mentioned method embodiments, which have similar implementation principles and technical effects, and will not be described here in detail.

[0190] In the above-mentioned embodiments, it should be understood that the processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC) and the like. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor and the like. The steps of the method disclosed in the present application can be directly embodied as the execution of the hardware processor, or executed by the combination of the hardware and software modules in the processor.

[0191] The memory can contain a random access memory (RAM), and can also include a non-volatile memory (NVM), for example, at least one disk memory.

[0192] The bus can be an industry standard architecture (ISA) bus, a peripheral component (PCI) bus or an extended industry standard architecture (EISA) bus and the like. The bus can be divided into an address bus, a data bus, a control bus and the like. For the convenience of representation, the bus in the drawings of the present application does not limit to only one bus or one type of bus.

[0193] ​The application further provides a computer program product comprising a computer program which, when executed by a processor, implements the method described above.

[0194] The application further provides a computer readable storage medium, wherein computer execution instructions are stored in the computer readable storage medium, and when a processor executes the computer execution instructions, the method described above is implemented.

[0195] The readable storage medium described above can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk. The readable storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0196] An exemplary readable storage medium is coupled to the processor, so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be an integral part of the processor. The processor and the readable storage medium can be located in an application specific integrated circuit (ASIC). Of course, the processor and the readable storage medium can also exist as discrete components in the device.

[0197] The division of units is only a logical function division, and in actual implementation, there can be another division mode, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0198] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or they can be distributed on a plurality of network units. According to actual needs, some or all of the units can be selected to achieve the purpose of the embodiment.

[0199] In addition, the functional units in each embodiment of the application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit.

[0200] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the parts of the technical solutions that essentially contribute to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the embodiments of the method of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0201] It can be understood by those skilled in the art that all or part of the steps of the above-mentioned method embodiments can be completed by program instruction related hardware. The aforementioned program can be stored in a computer readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk, and various media that can store program codes.

[0202] Finally, it should be noted that: those skilled in the art will easily think of other embodiments of the present application after considering the specification and practicing the application disclosed herein. The present application is intended to cover any variations, uses or adaptations of the present application that follow the general principles of the present application and include common knowledge or conventional technical means in the art that are not disclosed in the present application, and is not limited to the precise structure described above and shown in the drawings, and various modifications and changes can be made without departing from the scope thereof. The scope of the present application is only limited by the appended claims.

Claims

1. A data analysis method based on financial products, characterized in that: include: Responding to an analysis instruction from a user, obtaining basic information of a financial product to be analyzed; wherein the analysis instruction is used to instruct an analysis of the operating conditions of the financial product to be analyzed, and the basic information represents attributes of the financial product; Determine target feature information based on the basic information; wherein the target feature information represents the instructions for use of the financial product to be analyzed and the consumption data of the financial product during use; According to the target characteristic information, analysis result information is obtained; wherein the analysis result information represents the operation status of the financial product to be analyzed.

2. The method according to claim 1, characterized in that Determine target feature information based on the basic information, including: Determining first characteristic information and second characteristic information based on the basic information; wherein the first characteristic information represents the instructions for use of the financial product to be analyzed, and the second characteristic information represents interaction data during use of the financial product to be analyzed; The target feature information is determined according to the first feature information and the second feature information.

3. The method according to claim 2, characterized in that Determining first characteristic information based on the basic information includes: Based on a preset first association relationship, determining a preset document library corresponding to the basic information as a target document library; wherein the preset first association relationship represents an association relationship between the basic information and the preset document library, the preset document library includes a plurality of documents, and the documents represent instructions for use of the financial product to be analyzed; For each document in the target document library, determining a paragraph at a preset position in the document as summary information, and determining a first similarity between the summary information and the basic information; wherein the first similarity represents a degree of textual similarity between the summary information and the basic information; determining at least one target document from the target document library according to each of the first similarities; According to the at least one target document, the first feature information is obtained based on a preset first feature extraction network; wherein the preset first feature extraction network is used to input the target document and output the first feature information.

4. The method according to claim 2, characterized in that Determining second characteristic information based on the basic information includes: Based on a preset second association relationship, a preset interface set corresponding to the basic information is determined as a target interface set; wherein the preset second association relationship represents an association relationship between the basic information and the preset interface set, and the preset interface set includes a plurality of identification information, wherein the identification information is used to represent an interface, and the interface is used to query consumption data of the financial product corresponding to the interface during use; For each piece of identification information in the target interface set, determining a second similarity between the identification information and the basic information; wherein the second similarity represents a degree of textual similarity between the identification information and the basic information; Determining at least one target identification information from the target interface set based on each of the second similarities, and obtaining query information corresponding to the target identification information; wherein the query information represents consumption data of the financial product during use; According to the query information corresponding to the at least one target identification information, the second feature information is obtained based on a preset second feature extraction network; wherein the preset second feature extraction network is used to input the query information and output the second feature information.

5. The method according to any one of claims 1 to 4, characterized in that According to the target feature information, analysis result information is obtained, including: Determining an analysis text of the financial product to be analyzed based on the target feature information; Determining score information corresponding to the analyzed text; wherein the score information represents the degree of standardization of the analyzed text; If the score information meets the preset conditions, the analysis text is determined as the analysis result information of the financial product to be analyzed.

6. The method according to claim 5, characterized in that Determining the score information corresponding to the analyzed text includes: Determining a first scoring value, a second scoring value, and a third scoring value based on the analyzed text; wherein the first scoring value represents whether the analyzed text satisfies a preset rule, the second scoring value represents whether the description content of each sentence in the analyzed text is consistent, and the third scoring value represents whether preset keywords are covered in the analyzed text; The rating information is determined according to the first rating value, the second rating value, and the third rating value.

7. The method according to claim 6, characterized in that Determining a first scoring value according to the analyzed text includes: Determine the number of preset rules, which is a first number; For each preset rule, determining the logical state information corresponding to the preset rule based on the analysis text; wherein the logical state information is a first state or a second state, the first state indicating that the analysis text complies with the preset rule, and the second state indicating that the analysis text does not comply with the preset rule; Determine a second number based on the logic state information corresponding to each preset rule; wherein the second number represents the number of preset rules whose logic state information is in the first state; A ratio between the first number and the second number is determined as the first scoring value.

8. The method according to claim 6, characterized in that Determining a second scoring value according to the analyzed text includes: According to the analysis text, based on a preset word segmentation algorithm, determining combination information in the analysis text, and determining the quantity of the combination information, which is a third quantity; wherein the combination information includes field information and numerical information corresponding to the field information, the field information represents an attribute field contained in the analysis text, and the numerical information represents a numerical description corresponding to the attribute field; Clustering each combination information according to the field information in each combination information to obtain a combination set; wherein the field information of the combination information in each combination set is consistent; For each combination set, if there is consistent numerical information, determine the number of the numerical information in the combination set; Determining a fourth quantity based on the quantities corresponding to the respective numerical information in the combination set; wherein the fourth quantity is the maximum value among the quantities corresponding to the respective numerical information; A ratio between the third quantity and the fourth quantity is determined as the second scoring value.

9. The method according to claim 6, characterized in that Determining a third scoring value according to the result information includes: Determine the number of preset keywords, which is the fifth number; For each preset keyword, determining coverage status information corresponding to the preset keyword based on the analyzed text; wherein the coverage status information is a third state or a fourth state, the third state indicating that the analyzed text includes the preset keyword, and the fourth state indicating that the analyzed text does not include the preset keyword; Determining a sixth number based on the coverage status information corresponding to each preset keyword; wherein the sixth number represents the number of preset keywords whose coverage status information is in the third state; A ratio between the fifth quantity and the sixth quantity is determined as the third scoring value.

10. A data analysis device based on financial products, characterized in that: include: a response unit, configured to obtain basic information of a financial product to be analyzed in response to an analysis instruction from a user; wherein the analysis instruction is used to instruct an analysis of the operating conditions of the financial product to be analyzed, and the basic information represents attributes of the financial product; a determination unit, configured to determine target characteristic information based on the basic information; wherein the target characteristic information represents the instructions for use of the financial product to be analyzed and consumption data of the financial product during use; The analysis unit is used to obtain analysis result information based on the target feature information; wherein the analysis result information represents the operating status of the financial product to be analyzed.

Images