Information prediction method and device based on deep learning model, and electronic equipment

Abstract

The embodiment of the invention provides an information prediction method and device based on a deep learning model, and electronic equipment, and belongs to the technical field of financial science and technology. The method comprises the following steps: acquiring a plurality of financial materials, and performing information analysis on all the financial materials to obtain a feature set; selecting a separation value of the characteristic value through a decision tree algorithm, and dividing to obtain a first data set and a second data set; carrying out Gini coefficient calculation on the first data set and the second data set to obtain a target Gini coefficient; performing node screening on all financial materials according to the target Gini coefficient to construct a decision tree; performing node numbering on all leaf nodes in the decision tree to obtain a target material primary key; updating the financial material according to the target material primary key; and inputting the updated financial material to be predicted into the deep learning model, so that the deep learning model carries out information prediction based on the target material primary key, and a prediction result is displayed to a user. The information prediction accuracy can be improved, and the user experience is improved.

Description

Technical Field

[0001] The present application relates to the field of financial technology, and in particular to an information prediction method, device and electronic device based on a deep learning model. Background Art

[0002] In the digital age, the development of recommendation software and platforms has provided powerful technical support for processing and analyzing the interaction between users and materials. With the advancement of Internet technology, especially the development of big data and cloud computing, recommendation systems can process massive amounts of user behavior data and material information to provide users with personalized recommendation services. For tens of thousands of products, each product will be matched with different copywriting, introductions, and exhibition pictures. This results in the software or platform generating tens of thousands or hundreds of thousands of materials. Since these materials are used in different scenarios, such as popular scenarios such as shopping malls and coupons, the exposure and interaction of materials in these scenarios will be more and richer.

[0003] However, for other financial scenarios, such as bancassurance (bank-issued insurance), loans, etc., the proportion of materials and the probability of meeting customers are greatly reduced. For such material behavior data, the distribution is extremely uneven. Because of the small number of samples, the amount of information in the processing process is small, and even the vectors generated by training for some long-tail materials have large deviations and cannot truly reflect its relationship with other vectors in the entire vector space. This will lead to a reduction in the ability to predict materials, further reducing the accuracy of recommendations by recommendation software or recommendation platforms, and users may not be able to obtain financial product recommendations that match their needs and preferences, resulting in reduced satisfaction. Summary of the invention

[0004] The main purpose of the embodiments of the present application is to propose an information prediction method, device and electronic device based on a deep learning model, which can improve the accuracy of information prediction and improve user experience.

[0005] To achieve the above objectives, a first aspect of an embodiment of the present application proposes an information prediction method based on a deep learning model, the method comprising: Acquire multiple financial materials, and perform information analysis on all the financial materials to obtain a feature set of the financial materials, wherein the feature set includes multiple groups of feature values; For each set of the feature values ​​in the feature set, a separation value of the feature value is selected by a preset decision tree algorithm, and the financial materials are divided into materials according to the separation value to obtain a first data set and a second data set; Calculating the Gini coefficient for the first data set and the second data set to obtain a target Gini coefficient corresponding to the characteristic value; Performing node screening on all the financial materials according to the target Gini coefficient to construct a decision tree, wherein the decision tree includes a plurality of leaf nodes; Numbering all leaf nodes in the decision tree to obtain target material primary keys corresponding to the leaf nodes; Update the financial material according to the target material primary key to obtain the financial material to be predicted; The financial material to be predicted is input into a deep learning model, so that the deep learning model performs information prediction on the financial material to be predicted based on the target material primary key, obtains a prediction result, and displays the prediction result to a user.

[0006] In some embodiments, the information analysis of all the financial materials to obtain the feature set of the financial materials includes: Obtaining material identifications of all the financial materials, and determining the material identifications as material primary keys of the financial materials; Performing attribute analysis on the financial material to determine material attribute information of the financial material; Performing behavior analysis on the financial material to determine material behavior information of the financial material; Determining material characteristics of the financial material according to the material attribute information and the material behavior information; Binding the material characteristics and the material primary key to obtain multiple groups of characteristic values; A feature set is obtained according to the plurality of sets of feature values.

[0007] In some embodiments, dividing the financial materials according to the separation value to obtain the first data set and the second data set includes: Segmenting the material features in the financial material according to the separation value to determine positive samples and negative samples in the financial material, wherein the positive samples are samples that have been exposed and clicked, and the negative samples are samples that have been exposed but not clicked; A first data set is generated according to the positive samples, and a second data set is generated according to the negative samples.

[0008] In some embodiments, the calculating of the Gini coefficient for the first data set and the second data set to obtain a target Gini coefficient corresponding to the feature value includes: Obtaining the material quantity of the financial material, the number of positive samples in the first data set, and the number of negative samples in the second data set; Calculating a first Gini coefficient for the first data set, and calculating a second Gini coefficient for the second data set; Determine a first weight value of the first data set according to the number of materials and the number of positive samples, and determine a second weight value of the second data set according to the number of materials and the number of negative samples; Determine a first weighted value of the first data set according to the first weighted value and the first Gini coefficient, and determine a second weighted value of the second data set according to the second weighted value and the second Gini coefficient; An average operation is performed on the first weighted value and the second weighted value to obtain a weighted average Gini coefficient, and the weighted average Gini coefficient is determined as a target Gini coefficient corresponding to the characteristic value.

[0009] In some embodiments, the node screening of all the financial materials according to the target Gini coefficient to construct a decision tree includes: Determining an information gain value corresponding to the characteristic value in the financial material according to the target Gini coefficient; Screening the information gain value to obtain a target information gain value; Determine a target eigenvalue corresponding to the target information gain value among the eigenvalues, and use a node corresponding to the target eigenvalue as a root node; Calculating node information gain values ​​of feature values ​​other than the target feature value in the feature set until the feature set is traversed to obtain a node set; A decision tree is constructed according to the root node and the node set.

[0010] In some embodiments, the step of numbering all leaf nodes in the decision tree to obtain a target material primary key corresponding to the leaf node includes: All leaf nodes in the decision tree are numbered to obtain target node identifiers corresponding to the leaf nodes, and the target node identifiers are determined as target material primary keys of the leaf nodes.

[0011] In some embodiments, updating the financial material according to the target material primary key to obtain the financial material to be predicted includes: Performing node mapping on the target material primary key and the material primary key of the financial material to obtain a mapping set; Determine a target node corresponding to the target material primary key based on the mapping set, and determine the target material primary key of the target node; The target material primary key of the target node is updated to the target material primary key, and the financial material is updated according to the updated target node to obtain the predicted financial material.

[0012] To achieve the above-mentioned purpose, a second aspect of an embodiment of the present application proposes an information prediction device based on a deep learning model, the device comprising: A material acquisition module, used to acquire multiple financial materials, and perform information analysis on all the financial materials to obtain a feature set of the financial materials, wherein the feature set includes multiple groups of feature values; A material classification module, configured to select a separation value of the feature value for each set of feature values ​​in the feature set by a preset decision tree algorithm, and classify the financial materials according to the separation value to obtain a first data set and a second data set; A coefficient calculation module, used to calculate the Gini coefficient of the first data set and the second data set to obtain a target Gini coefficient corresponding to the characteristic value; A node screening module, used for screening nodes of all the financial materials according to the target Gini coefficient to construct a decision tree, wherein the decision tree includes a plurality of leaf nodes; A node numbering module, used to number all leaf nodes in the decision tree to obtain a target material primary key corresponding to the leaf node; A material update module, used for updating the financial material according to the target material primary key to obtain the financial material to be predicted; The information prediction module is used to input the financial material to be predicted into the deep learning model, so that the deep learning model performs information prediction on the financial material to be predicted based on the target material primary key, obtains the prediction result, and displays the prediction result to the user.

[0013] To achieve the above-mentioned objectives, the third aspect of an embodiment of the present application proposes an electronic device, an electronic device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the information prediction method based on the deep learning model as described in the first aspect when executing the computer program.

[0014] To achieve the above-mentioned purpose, the fourth aspect of an embodiment of the present application proposes a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the information prediction method based on the deep learning model as described in the first aspect.

[0015] The information prediction method, device, electronic device and storage medium based on the deep learning model proposed in the present application first obtain multiple financial materials, and perform information analysis on all financial materials to obtain a feature set of the financial materials. By performing information analysis on the financial materials, more abundant features can be obtained, so that the model can more accurately capture the interests and preferences of users. For each set of feature values ​​in the feature set, a separation value of the feature value is selected by a preset decision tree algorithm. The feature value is separated by selecting the separation value to reduce the complexity of calculation, and the financial materials are divided into materials according to the separation value to obtain a first data set and a second data set, so as to achieve accurate division of the data sets, facilitate subsequent finding of the optimal splitting point of the feature value, and improve the flexibility of information prediction. Then, the Gini coefficient is calculated for the first data set and the second data set, so as to measure the purity of the first data set and the second data set, and obtain the target Gini coefficient corresponding to the feature value. After that, all financial materials are node screened according to the target Gini coefficient, so as to screen out more important nodes and find the optimal splitting point of the feature value to create a decision tree, so that the material feature of the financial material can be divided into The target materials are divided into different leaf nodes, and then all the leaf nodes in the decision tree are numbered to obtain the new target material primary key corresponding to the leaf node, so that the position of the leaf node can be used as the primary key of the material characteristics in the financial material, and the material characteristics of the financial material can be generalized. Finally, the financial material is updated according to the target material primary key, so that the new target material primary key can be used as the new primary key of the material characteristics in the financial material, thereby avoiding the situation that the material characteristics with sparse behavior are not sufficiently trained in the deep network, and the financial material to be predicted is obtained, that is, the financial material used for information prediction in the future, so that the subsequent deep learning model can better capture the common characteristics of the material and improve the accuracy of information prediction. The financial material to be predicted is then input into the deep learning model, so that the deep learning model can perform information prediction on the financial material to be predicted based on the target material primary key. By performing information prediction on the updated financial material to be predicted, the accuracy of the click-through rate prediction of the sparse material can be enhanced, and the prediction result can be obtained, and the prediction result is displayed to the user, thereby improving the accuracy and reliability of the prediction, so that the user can obtain financial product recommendations that match their needs and preferences, and improve user satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a flowchart of an information prediction method based on a deep learning model provided in an embodiment of the present application; Figure 2 A flow chart of a method for performing information analysis on all financial materials provided in an embodiment of the present application; Figure 3 A flow chart of a method for classifying financial materials according to separation values ​​provided in an embodiment of the present application; Figure 4 yes Figure 1 Flow chart of step S103 in FIG. Figure 5 yes Figure 1 Flow chart of step S104 in FIG. Figure 6 yes Figure 1 Flow chart of step S105 in FIG. Figure 7 yes Figure 1 Flow chart of step S106 in FIG. Figure 8 is a schematic diagram of the structure of an information prediction device based on a deep learning model provided in an embodiment of the present application; Fig. 9 It is a schematic diagram of the hardware structure of the electronic device provided in the embodiment of the present application. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0018] It should be noted that, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second", etc. in the specification, claims and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0020] First, some nouns involved in this application are analyzed: Natural Language Processing (NLP): NLP uses computers to process, understand, and apply human languages ​​(such as Chinese, English, etc.). NLP is a branch of artificial intelligence and an interdisciplinary subject between computer science and linguistics. It is often referred to as computational linguistics. Natural language processing includes grammatical analysis, semantic analysis, and text understanding. Natural language processing is commonly used in technical fields such as machine translation, handwritten and printed character recognition, speech recognition and text-to-speech conversion, information intent recognition, information extraction and filtering, text classification and clustering, public opinion analysis, and opinion mining. It involves data mining related to language processing, machine learning, knowledge acquisition, knowledge engineering, artificial intelligence research, and linguistic research related to language computing.

[0021] Classification and Regression Trees (CART): CART is a decision tree algorithm for classification and regression. CART builds a model by recursively partitioning the feature space into smaller and smaller regions, each of which is associated with one or more decision rules.

[0022] Iterative Dichotomiser 3 (ID3): It is an early decision tree algorithm mainly used for classification problems. The ID3 algorithm uses information gain as the criterion for feature selection. It starts with the training data set, selects the feature with the largest information gain as the node for splitting, and recursively builds a decision tree until the stopping condition is met, such as all instances belong to the same class or no more features are available.

[0023] Gini Coefficient (Gini): An indicator proposed by Italian economist Gini to quantitatively measure the degree of income distribution differences. It is an important analytical indicator used internationally to comprehensively examine the income distribution differences among residents.

[0024] The information prediction method and device based on the deep learning model, electronic device and storage medium provided in the embodiments of the present application can improve the accuracy of information prediction and enhance user experience.

[0025] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0026] The basic technologies of artificial intelligence generally include sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. Artificial intelligence software technologies mainly include computer vision technology, robotics technology, biometrics technology, module management technology of online meeting systems, natural language processing technology, and machine learning / deep learning.

[0027] The information prediction method based on deep learning model provided in the embodiment of the present application relates to the field of financial technology. The information prediction method based on deep learning model provided in the embodiment of the present application can be applied to the terminal, can also be applied to the server side, and can also be software running in the terminal or the server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or a server cluster or distributed system composed of multiple physical servers, and can also be configured as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the information prediction method based on the deep learning model, etc., but is not limited to the above forms.

[0028] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0029] It should be noted that in each specific implementation of the present application, when it comes to the need to perform relevant processing based on data related to user identity or characteristics such as user information, user behavior data, user historical data, and user location information, the user's permission or consent will be obtained first, and the collection, use, and processing of these data will comply with the relevant laws, regulations, and standards of the relevant countries and regions. In addition, when the embodiment of the present application needs to obtain the user's sensitive personal information, the user's separate permission or consent will be obtained through a pop-up window or by jumping to a confirmation page. After clearly obtaining the user's separate permission or consent, the necessary user-related data for the normal operation of the embodiment of the present application will be obtained.

[0030] In the digital age, the development of recommendation software and platforms has provided powerful technical support for processing and analyzing the interaction between users and materials. With the advancement of Internet technology, especially the development of big data and cloud computing, recommendation systems can process massive amounts of user behavior data and material information to provide users with personalized recommendation services. For tens of thousands of products, each product will be matched with different copywriting, introductions, and exhibition pictures. This results in the software or platform generating tens of thousands or hundreds of thousands of materials. Since these materials are used in different scenarios, such as popular scenarios such as shopping malls and coupons, the exposure and interaction of materials in these scenarios will be more and richer.

[0031] However, for other financial scenarios, such as bancassurance (bank-issued insurance), loans, etc., the proportion of materials and the probability of meeting customers are greatly reduced. For such material behavior data, the distribution is extremely uneven. Because of the small number of samples, the amount of information in the processing process is small, and even the vectors generated by training for some long-tail materials have large deviations and cannot truly reflect its relationship with other vectors in the entire vector space. This will lead to a reduction in the ability to predict materials, further reducing the accuracy of recommendations by recommendation software or recommendation platforms, and users may not be able to obtain financial product recommendations that match their needs and preferences, resulting in reduced satisfaction.

[0032] In order to solve the above problems, the present embodiment provides an information prediction method, device, electronic device and storage medium based on a deep learning model. First, multiple financial materials are obtained, and information analysis is performed on all financial materials to obtain a feature set of the financial materials. By performing information analysis on the financial materials, more abundant features can be obtained, so that the model can more accurately capture the interests and preferences of users. For each set of feature values ​​in the feature set, a separation value of the feature value is selected by a preset decision tree algorithm. The feature value is separated by selecting the separation value to reduce the complexity of calculation. The financial materials are divided into materials according to the separation value to obtain a first data set and a second data set, so as to achieve accurate division of the data sets, facilitate subsequent finding of the optimal splitting point of the feature value, and improve the flexibility of information prediction. The Gini coefficient is then calculated for the first data set and the second data set, so as to measure the purity of the first data set and the second data set, and obtain the target Gini coefficient corresponding to the feature value. After that, nodes of all financial materials are screened according to the target Gini coefficient, so as to screen out more important nodes and find the optimal splitting point of the feature value to create a decision tree, so as to be able to divide the financial materials. The material features are divided into different leaf nodes, and then all the leaf nodes in the decision tree are numbered to obtain the new target material primary key corresponding to the leaf node, so that the position of the leaf node can be used as the primary key of the material features in the financial material, and the material features of the financial material can be generalized. Finally, the financial material is updated according to the target material primary key, so that the new target material primary key can be used as the new primary key of the material features in the financial material, thereby avoiding the situation that the material features with sparse behavior are not sufficiently trained in the deep network, and the financial material to be predicted is obtained, that is, the financial material used for information prediction in the future, so that the subsequent deep learning model can better capture the common characteristics of the material and improve the accuracy of information prediction. The financial material to be predicted is then input into the deep learning model, so that the deep learning model performs information prediction on the financial material to be predicted based on the target material primary key. By performing information prediction on the updated financial material to be predicted, the accuracy of the click-through rate prediction of the sparse material can be enhanced, and the prediction result can be obtained and displayed to the user, thereby improving the accuracy and reliability of the prediction, so that the user can obtain financial product recommendations that match their needs and preferences, and improve user satisfaction.

[0033] The following is a detailed description with reference to the accompanying drawings.

[0034] Figure 1 is an optional flowchart of the information prediction method based on the deep learning model provided in the embodiment of the present application. Figure 1 The method may include but is not limited to steps S101 to S107.

[0035] Step S101, obtaining multiple financial materials, and performing information analysis on all the financial materials to obtain a feature set of the financial materials, wherein the feature set includes multiple groups of feature values.

[0036] It should be noted that the characteristic value is used to characterize the correspondence between the material characteristics and the material primary key in the financial material.

[0037] In step S101 of some embodiments, the embodiments of the present application first obtain multiple financial materials. Specifically, the financial materials can be obtained from an open source database or user historical click data. The financial materials include but are not limited to stocks, bonds, transaction prices, trading volumes and other information, and information analysis is performed on all financial materials to obtain material attribute information and material behavior information of the financial materials, and further obtain a feature set to provide richer features for the model, wherein the feature values ​​in the feature set are used to characterize the correspondence between the material features in the financial materials and the material primary keys, so as to more accurately capture the user's interests and preferences and provide the user with more personalized recommendations.

[0038] It should be noted that the open source database can be obtained through channels such as transaction data, market situation database, financial information service platform, etc., and the embodiments of this application do not impose specific restrictions.

[0039] Step S102: for each set of feature values ​​in the feature set, a separation value of the feature value is selected by a preset decision tree algorithm, and the financial materials are divided into materials according to the separation value to obtain a first data set and a second data set.

[0040] In step S102 of some embodiments, for each set of feature values ​​in the feature set, a separation value of the feature value is selected by a preset decision tree algorithm. Specifically, for categorical features, the separation value in the embodiment of the present application may be a specific category of the feature; for continuous features, the separation value in the embodiment of the present application may be a value within the range of the feature value. The financial materials are divided according to the separation value to obtain the first data set and the second data set, thereby achieving accurate separation of the data sets, and being able to flexibly separate according to different types of features, thereby improving the flexibility of separation.

[0041] It should be noted that the decision tree algorithm in the embodiment of the present application may be an ID3 algorithm, a CART algorithm, etc. Specifically, the embodiment of the present application uses the CART decision tree binary classification algorithm to find the optimal splitting point of the eigenvalue.

[0042] Step S103, calculating the Gini coefficients of the first data set and the second data set to obtain a target Gini coefficient corresponding to the characteristic value.

[0043] In step S103 of some embodiments, the Gini coefficient is calculated for the first data set and the second data set, so as to obtain the Gini coefficient of each data set. The purity of the first data set and the second data set can be further measured by the Gini coefficient to obtain the target Gini coefficient corresponding to the feature value, thereby obtaining the Gini coefficient of a certain feature after splitting according to a certain separation value, so that the model can classify the data more effectively, further improving the prediction accuracy of the model.

[0044] Step S104, performing node screening on all financial materials according to the target Gini coefficient to construct a decision tree, wherein the decision tree includes a plurality of leaf nodes.

[0045] In step S104 of some embodiments, nodes of all financial materials are screened according to the target Gini coefficient, which can reduce the overfitting of the model to the training data and improve the generalization ability of the model on unknown data, so as to construct a decision tree so that each material primary key falls on a leaf node, ensuring that each material primary key has a corresponding leaf node, which facilitates the subsequent generalization of the material primary key and avoids the problem of poor prediction results and low accuracy due to the small amount of information contained in the vector.

[0046] Step S105, numbering all leaf nodes in the decision tree to obtain the target material primary key corresponding to the leaf node.

[0047] In step S105 of some embodiments, all leaf nodes in the decision tree are numbered to obtain the target material primary key corresponding to the leaf node, so that a new target material primary key can be generated for each leaf node, and the magnitude of the new target material primary key is smaller than the magnitude of the original material primary key, which facilitates subsequent data integration, reduces storage requirements, and reduces the overhead of deep model identification training for material features.

[0048] Step S106, updating the financial material according to the target material primary key to obtain the financial material to be predicted.

[0049] In step S106 of some embodiments, the financial material is updated according to the target material primary key to obtain the financial material to be predicted, thereby alleviating the problem of insufficient training of sparse behavior identifiers in the deep network.

[0050] Step S107, input the financial material to be predicted into the deep learning model, so that the deep learning model performs information prediction on the financial material to be predicted based on the primary key of the target material, obtains the prediction result, and displays the prediction result to the user.

[0051] In step S107 of some embodiments, the financial material to be predicted is input into the deep learning model, so that the deep learning model performs information prediction on the financial material to be predicted based on the primary key of the target material, thereby enhancing the accuracy of the click-through rate prediction of sparse materials, obtaining the prediction results, and displaying the prediction results to the user, and being able to recommend personalized information to the user, so that the user can obtain financial product recommendations that match their needs and preferences, thereby improving user satisfaction.

[0052] It should be noted that in the process of the deep learning model performing information prediction on the financial material to be predicted based on the target material primary key, the embodiment of the present application will use the newly generated target material primary key instead of the material primary key as the input of the sparse features of the neural network in the deep learning model, thereby alleviating the problem of insufficient training of sparse behavior identifiers in the deep network and improving the accuracy of information prediction.

[0053] See also Figure 2 , Figure 2 A flow chart of a method for performing information analysis on all financial materials provided in an embodiment of the present application, the method including but not limited to steps S201 to S206.

[0054] Step S201, obtaining the material identification of all financial materials, and determining the material identification as the material primary key of the financial material.

[0055] Step S202: Analyze the attributes of the financial materials to determine the material attribute information of the financial materials.

[0056] Step S203: Performing behavior analysis on the financial material to determine material behavior information of the financial material.

[0057] Step S204: determining the material characteristics of the financial material according to the material attribute information and the material behavior information.

[0058] Step S205, binding the material characteristics and the material primary key to obtain multiple groups of characteristic values.

[0059] Step S206, obtaining a feature set according to the multiple sets of feature values.

[0060] In steps S201 to S206 of some embodiments, in the process of performing information analysis on all financial materials, the embodiments of the present application first obtain the material identification of all financial materials, wherein the material identification of each financial material is unique, and determine the material identification as the material primary key of the financial material. The material primary key, as the unique identification of the material, ensures that each financial material can be accurately identified and distinguished, and the material primary key can simplify the query and retrieval process, improve the work efficiency of financial institutions when processing a large number of financial materials, and then perform attribute analysis on the financial materials to determine the material attribute information of the financial materials, such as the specifications, costs, suppliers, etc. of the materials, thereby providing a richer model. Features, then conduct behavioral analysis on financial materials to determine the material behavior information of financial materials, such as the exposure rate and click-through rate of financial materials in a certain time window. By analyzing the behavioral data of materials, users' interests and preferences can be captured more accurately, and the material characteristics of financial materials can be determined based on the material attribute information and material behavior information, which can provide users with more personalized recommendations. The material characteristics and material primary keys are then bound to obtain multiple sets of feature values, and then feature sets are obtained based on the multiple sets of feature values. The material primary key is the unique identifier of the material, and binding it with the material characteristics can ensure the consistency and accuracy of the material information, reduce data errors and redundancy, and improve data reliability.

[0061] It should be noted that, in the embodiment of the present application, the material characteristics include material attribute information and material behavior information such as exposure and click of financial materials in a certain time window, and the material identification of all financial materials is obtained, and the material identification is determined as the material primary key of the financial material, which is recorded as ,… , and then record the material characteristics as , ,.... , the corresponding relationship between the material primary key and the material characteristics is :{ , ,.... }; :{ , ,.... }.

[0062] See also Figure 3 , Figure 3 A flow chart of a method for classifying financial materials according to separation values ​​provided in an embodiment of the present application, the method includes but is not limited to steps S301 to S302.

[0063] Step S301, segmenting material features in the financial material according to the separation value, and determining positive samples and negative samples in the financial material.

[0064] It should be noted that positive samples are samples that have been exposed and clicked, and negative samples are samples that have been exposed but not clicked. Specifically, when a financial material is displayed to a user and the user clicks on it, the clicked financial material is regarded as a positive sample; when a financial material is displayed to a user and the user does not click on it, the displayed financial material is regarded as a negative sample.

[0065] Step S302: Generate a first data set based on positive samples, and generate a second data set based on negative samples.

[0066] In steps S301 to S302 of some embodiments, in the process of dividing the financial materials according to the separation value, the embodiment of the present application first divides the material features in the financial materials according to the separation value, and determines the positive samples and negative samples in the financial materials. Specifically, taking the separation value in the embodiment of the present application as a specific category as an example, the separation value of the material that has been exposed but not clicked is set to 0, and the separation value of the material that has been exposed and clicked is set to 1. The positive samples that have been exposed and clicked and the negative samples that have been exposed and not clicked are screened out through the separation value, and then a first data set is generated according to the positive samples, and a second data set is generated according to the negative samples. Through the clear separation value, accurate separation of the data sets is achieved, and flexible separation can be performed according to different types of features, thereby improving the flexibility of segmentation.

[0067] See also Figure 4 In some embodiments, step S103 may also include but is not limited to steps S401 to S405.

[0068] Step S401, obtaining the material quantity of financial materials, the number of positive samples in the first data set, and the number of negative samples in the second data set.

[0069] Step S402: Calculate a first Gini coefficient for the first data set, and calculate a second Gini coefficient for the second data set.

[0070] Step S403, determining a first weight value of the first data set according to the number of materials and the number of positive samples, and determining a second weight value of the second data set according to the number of materials and the number of negative samples.

[0071] Step S404: determining a first weighted value of the first data set according to the first weighted value and the first Gini coefficient, and determining a second weighted value of the second data set according to the second weighted value and the second Gini coefficient.

[0072] Step S405: average the first weighted value and the second weighted value to obtain a weighted average Gini coefficient, and determine the weighted average Gini coefficient as a target Gini coefficient corresponding to the characteristic value.

[0073] In steps S401 to S405 of some embodiments, in the process of calculating the Gini coefficient for the first data set and the second data set, the embodiment of the present application first obtains the material quantity of the financial material, the number of positive samples in the first data set, and the number of negative samples in the second data set, so as to facilitate the subsequent determination of the proportion of the first data set and the second data set in the material quantity. Then the first Gini coefficient of the first data set is calculated, and the second Gini coefficient of the second data set is calculated, so that the Gini coefficient of each data set can be obtained, and the purity of the first data set and the second data set can be further measured by the Gini coefficient. Then, the first weight value of the first data set is determined according to the material quantity and the number of positive samples, and the second weight value of the second data set is determined according to the material quantity and the number of negative samples. Specifically, the proportion of the number of positive samples in the material quantity is calculated to obtain the first weight value, and the proportion of negative sample data in the material data is calculated to obtain the second weight value, so that the proportion of positive samples and negative samples in the total material quantity can be obtained. After that, the first weight value of the first data set is determined according to the first weight value and the first Gini coefficient, and the second weight value of the second data set is determined according to the first weight value and the first Gini coefficient. The second weight value and the second Gini coefficient determine the second weight value of the second data set. Specifically, the first weight value is multiplied by the first Gini coefficient to obtain the first weight value. Similarly, the second weight value is multiplied by the second Gini coefficient to obtain the second weight value, that is, the Gini coefficient of each sample is multiplied by its weight. Finally, the first weight value and the second weight value are averaged, that is, the first weight value and the second weight value are summed and then the sum is divided by the sum of all weight values ​​to obtain the weighted average Gini coefficient, and the weighted average Gini coefficient is determined as the target Gini coefficient corresponding to the feature value, so as to obtain the Gini coefficient of a certain feature after splitting according to a certain separation value, so that the model can classify the data more effectively and further improve the prediction accuracy of the model.

[0074] It should be noted that the Gini coefficient is an indicator used to measure the purity of a data set. The smaller the value, the higher the purity of the data set, that is, the samples in the data set tend to belong to the same category. In the process of calculating the first Gini coefficient of the first data set and calculating the second Gini coefficient of the second data set, the embodiment of the present application will first calculate the first mathematical expectation of the positive samples in the first data set and the second mathematical expectation of the negative samples in the second data set. Specifically, the formula for calculating the Gini coefficient is as follows: G(p)= =1- ; in, is the mathematical expectation of positive or negative samples.

[0075] The formula for calculating the weighted average Gini coefficient is as follows: G(D, )= *G( ); Among them, G( ) is the Gini coefficient, that is, the first Gini coefficient or the second Gini coefficient in the embodiment of the present application, is the number of sub-datasets, The material quantity for financial materials.

[0076] See also Figure 5 In some embodiments, step S104 may also include but is not limited to steps S501 to S505.

[0077] Step S501, determining the information gain value corresponding to the characteristic value in the financial material according to the target Gini coefficient.

[0078] Step S502: Screen the information gain values ​​to obtain a target information gain value.

[0079] Step S503: determine a target eigenvalue corresponding to the target information gain value in the eigenvalues, and use the node corresponding to the target eigenvalue as the root node.

[0080] In steps S501 to S503 of some embodiments, in the process of node screening of all financial materials according to the target Gini coefficient, the embodiment of the present application first determines the information gain value corresponding to the characteristic value in the financial material according to the target Gini coefficient, so that the splitting properties of different features in the financial material can be judged by the information gain value, which provides a key indicator for node selection. Specifically, the smaller the target Gini coefficient in the embodiment of the present application, the larger the information gain value brought. Then, the information gain value is screened, and the maximum information gain value is used as the target information gain value. After that, the target characteristic value corresponding to the target information gain value is determined in the characteristic value, and the node corresponding to the target characteristic value is used as the root node, that is, the material feature corresponding to the target characteristic value and its separation value are used as the root node, and the optimal splitting point of the characteristic value is found. By selecting characteristic values ​​with high information gain, the overfitting of the model to the training data can be reduced, and the generalization ability of the model on unknown data can be improved. Moreover, characteristic values ​​with high information gain can better reduce the uncertainty of the data set. Therefore, in the construction of the decision tree, selecting features with large information gain as splitting nodes can improve the classification ability of the model.

[0081] Step S504, calculating the node information gain values ​​of the feature values ​​other than the target feature value in the feature set, until the feature set is traversed to obtain a node set.

[0082] Step S505: construct a decision tree according to the root node and the node set.

[0083] In steps S504 to S505 of some embodiments, after determining the root node, the embodiments of the present application will also calculate the node information gain values ​​of the feature values ​​other than the target feature value in the feature set. Specifically, continue to calculate the splitting point of a certain material feature that brings the largest information gain value, and then use the material feature and its separation value as the root node, repeat step S102 and steps S501 to S503 until the feature set is traversed to obtain a node set. Finally, a decision tree is constructed based on the root node and the node set, so that each material primary key will fall on a leaf node, ensuring that each material primary key will have a corresponding leaf node, which is convenient for the subsequent generalization of the material primary key and avoids the problem of poor prediction results and low accuracy due to the small amount of information contained in the model vector.

[0084] See also Figure 6 In some embodiments, step S105 may also include but is not limited to step S601.

[0085] Step S601: number all leaf nodes in the decision tree, obtain target node identifiers corresponding to the leaf nodes, and determine the target node identifiers as target material primary keys of the leaf nodes.

[0086] In step S601 of some embodiments, in the process of numbering all leaf nodes in the decision tree, the embodiments of the present application sequentially number all leaf nodes in the decision tree, for example, numbered as ,… , obtain the target node identifier corresponding to the leaf node, and determine the target node identifier as the target material primary key of the leaf node, so that a new target material primary key can be generated for each leaf node, and the magnitude of the new target material primary key is smaller than the magnitude of the original material primary key, which is convenient for subsequent data integration, can reduce storage requirements, and can reduce the overhead of deep model identification training for material features.

[0087] It should be noted that the embodiment of the present application does not need to prune the decision tree, and the risk of overfitting of the decision tree can be reduced by controlling the quality and quantity of material features, for example, controlling the number of material features between 10 and 30, and using features with sufficiently high saturation. And it can ensure that each will have a corresponding leaf node number.

[0088] See also Figure 7 In some embodiments, step S106 may also include but is not limited to steps S701 to S703.

[0089] Step S701, perform node mapping on the target material primary key and the material primary key of the financial material to obtain a mapping set.

[0090] Step S702: determine the target node corresponding to the target material primary key based on the mapping set, and determine the target material primary key of the target node.

[0091] Step S703: update the target material primary key of the target node to the target material primary key, and update the financial material according to the updated target node to obtain the predicted financial material.

[0092] In steps S701 to S703 of some embodiments, in the process of updating the financial material according to the target material primary key, the embodiment of the present application first performs node mapping on the target material primary key and the material primary key of the financial material to obtain a mapping set, so as to obtain the corresponding relationship between the target material primary key and the material primary key, thereby ensuring the consistency of the data and ensuring that the target material primary key and the material primary key of the same node correspond to each other, wherein the mapping relationship can be expressed as { : }, then, based on the mapping set, determine the target node corresponding to the target material primary key, and determine the target material primary key of the target node. By mapping the set, errors caused by manual matching can be reduced. Finally, the target material primary key of the target node is updated to the target material primary key, that is, the newly generated replace , and update the financial material according to the updated target node to obtain the predicted financial material, which can alleviate the problem of insufficient training of sparse behavior identifiers in deep networks, thereby enhancing the accuracy of sparse material click-through rate prediction and recommending personalized information to users.

[0093] See also Figure 8 The embodiment of the present application also provides an information prediction device based on a deep learning model, the device comprising: The material acquisition module 801 is used to acquire multiple financial materials and perform information analysis on all the financial materials to obtain a feature set of the financial materials, wherein the feature set includes multiple sets of feature values; A material classification module 802 is used to select a separation value of the feature value for each set of feature values ​​in the feature set through a preset decision tree algorithm, and classify the financial materials according to the separation value to obtain a first data set and a second data set; A coefficient calculation module 803 is used to calculate the Gini coefficient of the first data set and the second data set to obtain a target Gini coefficient corresponding to the characteristic value; A node screening module 804 is used to screen nodes of all financial materials according to a target Gini coefficient to construct a decision tree, wherein the decision tree includes a plurality of leaf nodes; The node numbering module 805 is used to number all leaf nodes in the decision tree to obtain the target material primary key corresponding to the leaf node; The material update module 806 is used to update the financial material according to the target material primary key to obtain the financial material to be predicted; The information prediction module 807 is used to input the financial material to be predicted into the deep learning model, so that the deep learning model performs information prediction on the financial material to be predicted based on the primary key of the target material, obtains the prediction result, and displays the prediction result to the user.

[0094] The specific implementation of the information prediction device based on the deep learning model is basically the same as the specific implementation of the information prediction method based on the deep learning model mentioned above, and will not be repeated here.

[0095] The embodiment of the present application also provides an electronic device, the electronic device comprising: a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for realizing connection and communication between the processor and the memory, and the program is executed by the processor to realize the above-mentioned information prediction method based on the deep learning model. The electronic device can be any intelligent terminal including a tablet computer, a car computer, etc.

[0096] See also Fig. 9 , Fig. 9 The hardware structure of an electronic device of another embodiment is illustrated, and the electronic device includes: The processor 901 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application; The memory 902 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 902, and the processor 901 calls and executes the information prediction method based on the deep learning model in the embodiment of this application; Input / output interface 903, used to implement information input and output; Communication interface 904, used to realize communication interaction between the device and other devices, which can be realized through wired mode (such as USB, network cable, etc.) or wireless mode (such as mobile network, WIFI, Bluetooth, etc.); A bus 905 that transmits information between various components of the device (e.g., the processor 901, the memory 902, the input / output interface 903, and the communication interface 904); The processor 901 , the memory 902 , the input / output interface 903 and the communication interface 904 are connected to each other in communication within the device via a bus 905 .

[0097] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the above-mentioned information prediction method based on the deep learning model.

[0098] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0099] The information prediction method, device, electronic device and storage medium based on the deep learning model provided by the embodiments of the present application first obtain multiple financial materials, and perform information analysis on all the financial materials to obtain a feature set of the financial materials. By performing information analysis on the financial materials, more abundant features can be obtained, so that the model can more accurately capture the interests and preferences of users. For each set of feature values ​​in the feature set, a separation value of the feature value is selected by a preset decision tree algorithm. The feature value is separated by selecting the separation value, thereby reducing the complexity of calculation. The financial materials are divided into materials according to the separation value to obtain a first data set and a second data set, thereby achieving accurate division of the data sets, facilitating the subsequent finding of the optimal splitting point of the feature value, and improving the flexibility of information prediction. The Gini coefficient is then calculated for the first data set and the second data set, thereby measuring the purity of the first data set and the second data set, and obtaining a target Gini coefficient corresponding to the feature value. Afterwards, nodes of all financial materials are screened according to the target Gini coefficient, thereby screening out relatively important nodes, finding the optimal splitting point of the feature value, and creating a decision tree, thereby being able to divide the material characteristics of the financial materials. The characteristics are divided into different leaf nodes, and then all the leaf nodes in the decision tree are numbered to obtain the new target material primary key corresponding to the leaf node, so that the position of the leaf node can be used as the primary key of the material characteristics in the financial material, and the material characteristics of the financial material can be generalized. Finally, the financial material is updated according to the target material primary key, so that the new target material primary key can be used as the new primary key of the material characteristics in the financial material, thereby avoiding the situation that the material characteristics with sparse behavior are not sufficiently trained in the deep network, and the financial material to be predicted is obtained, that is, the financial material used for information prediction in the future, so that the subsequent deep learning model can better capture the common characteristics of the material and improve the accuracy of information prediction. The financial material to be predicted is then input into the deep learning model, so that the deep learning model performs information prediction on the financial material to be predicted based on the target material primary key. By performing information prediction on the updated financial material to be predicted, the accuracy of the click-through rate prediction of the sparse material can be enhanced, and the prediction result can be obtained, and the prediction result is displayed to the user, thereby improving the accuracy and reliability of the prediction, so that the user can obtain financial product recommendations that match their needs and preferences, and improve user satisfaction.

[0100] The embodiments described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0101] It can be understood by those skilled in the art that Figure 1-9The technical solutions shown in the figure do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than those shown in the figure, or a combination of certain steps, or different steps.

[0102] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0103] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0104] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0105] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0106] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0107] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0108] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0109] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including multiple instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store programs.

[0110] The preferred embodiments of the present application are described above with reference to the accompanying drawings, but the scope of the rights of the present application is not limited thereto. Any modification, equivalent substitution and improvement made by a person skilled in the art without departing from the scope and essence of the present application should be within the scope of the rights of the present application.

Claims

1. An information prediction method based on a deep learning model, characterized in that: The method comprises: Acquire multiple financial materials, and perform information analysis on all the financial materials to obtain a feature set of the financial materials, wherein the feature set includes multiple groups of feature values; For each set of the feature values ​​in the feature set, a separation value of the feature value is selected by a preset decision tree algorithm, and the financial materials are divided into materials according to the separation value to obtain a first data set and a second data set; Calculating the Gini coefficient for the first data set and the second data set to obtain a target Gini coefficient corresponding to the characteristic value; Performing node screening on all the financial materials according to the target Gini coefficient to construct a decision tree, wherein the decision tree includes a plurality of leaf nodes; Numbering all leaf nodes in the decision tree to obtain target material primary keys corresponding to the leaf nodes; Update the financial material according to the target material primary key to obtain the financial material to be predicted; The financial material to be predicted is input into a deep learning model, so that the deep learning model performs information prediction on the financial material to be predicted based on the target material primary key, obtains a prediction result, and displays the prediction result to a user.

2. The information prediction method based on deep learning model according to claim 1, characterized in that: The information analysis of all the financial materials is performed to obtain a feature set of the financial materials, including: Obtaining material identifications of all the financial materials, and determining the material identifications as material primary keys of the financial materials; Performing attribute analysis on the financial material to determine material attribute information of the financial material; Performing behavior analysis on the financial material to determine material behavior information of the financial material; Determining material characteristics of the financial material according to the material attribute information and the material behavior information; Binding the material characteristics and the material primary key to obtain multiple groups of characteristic values; A feature set is obtained according to the plurality of sets of feature values.

3. The information prediction method based on deep learning model according to claim 2, characterized in that: The material classification of the financial materials according to the separation value to obtain the first data set and the second data set includes: Segmenting the material features in the financial material according to the separation value to determine positive samples and negative samples in the financial material, wherein the positive samples are samples that have been exposed and clicked, and the negative samples are samples that have been exposed but not clicked; A first data set is generated according to the positive samples, and a second data set is generated according to the negative samples.

4. The information prediction method based on deep learning model according to claim 3, characterized in that: The calculating the Gini coefficient of the first data set and the second data set to obtain a target Gini coefficient corresponding to the characteristic value includes: Obtaining the material quantity of the financial material, the number of positive samples in the first data set, and the number of negative samples in the second data set; Calculating a first Gini coefficient for the first data set, and calculating a second Gini coefficient for the second data set; Determine a first weight value of the first data set according to the number of materials and the number of positive samples, and determine a second weight value of the second data set according to the number of materials and the number of negative samples; Determine a first weighted value of the first data set according to the first weighted value and the first Gini coefficient, and determine a second weighted value of the second data set according to the second weighted value and the second Gini coefficient; An average operation is performed on the first weighted value and the second weighted value to obtain a weighted average Gini coefficient, and the weighted average Gini coefficient is determined as a target Gini coefficient corresponding to the characteristic value.

5. The information prediction method based on deep learning model according to claim 1, characterized in that: The node screening of all the financial materials according to the target Gini coefficient to construct a decision tree includes: Determining an information gain value corresponding to the characteristic value in the financial material according to the target Gini coefficient; Screening the information gain value to obtain a target information gain value; Determine a target eigenvalue corresponding to the target information gain value among the eigenvalues, and use a node corresponding to the target eigenvalue as a root node; Calculating node information gain values ​​of feature values ​​other than the target feature value in the feature set until the feature set is traversed to obtain a node set; A decision tree is constructed according to the root node and the node set.

6. The information prediction method based on deep learning model according to claim 1, characterized in that: The step of numbering all leaf nodes in the decision tree to obtain the target material primary key corresponding to the leaf node includes: All leaf nodes in the decision tree are numbered to obtain target node identifiers corresponding to the leaf nodes, and the target node identifiers are determined as target material primary keys of the leaf nodes.

7. The information prediction method based on deep learning model according to claim 6, characterized in that: The updating of the financial material according to the target material primary key to obtain the financial material to be predicted includes: Performing node mapping on the target material primary key and the material primary key of the financial material to obtain a mapping set; Determine a target node corresponding to the target material primary key based on the mapping set, and determine the target material primary key of the target node; The target material primary key of the target node is updated to the target material primary key, and the financial material is updated according to the updated target node to obtain the predicted financial material.

8. An information prediction device based on a deep learning model, characterized in that: The device comprises: A material acquisition module, used to acquire multiple financial materials, and perform information analysis on all the financial materials to obtain a feature set of the financial materials, wherein the feature set includes multiple groups of feature values; A material classification module, configured to select a separation value of the feature value for each set of feature values ​​in the feature set by a preset decision tree algorithm, and classify the financial materials according to the separation value to obtain a first data set and a second data set; A coefficient calculation module, used to calculate the Gini coefficient of the first data set and the second data set to obtain a target Gini coefficient corresponding to the characteristic value; A node screening module, used for screening nodes of all the financial materials according to the target Gini coefficient to construct a decision tree, wherein the decision tree includes a plurality of leaf nodes; A node numbering module, used to number all leaf nodes in the decision tree to obtain a target material primary key corresponding to the leaf node; A material update module, used for updating the financial material according to the target material primary key to obtain the financial material to be predicted; The information prediction module is used to input the financial material to be predicted into the deep learning model, so that the deep learning model performs information prediction on the financial material to be predicted based on the target material primary key, obtains the prediction result, and displays the prediction result to the user.

9. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the information prediction method based on the deep learning model described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the information prediction method based on a deep learning model described in any one of claims 1 to 7 is implemented.

Images