Data verification method and device, equipment, storage medium and product

Abstract

The invention relates to the technical field of data processing, and particularly discloses a data verification method and device, equipment, a storage medium and a product. The method comprises the steps of obtaining to-be-verified business data of a target mechanism at a target time point, a target internal prediction model set and a target external prediction model set; on the basis of the to-be-verified business data and the verification passing data of each similar mechanism, predicting the target time point according to the target internal prediction model set and the target external prediction model set, and obtaining a corresponding internal modeling prediction result and a corresponding external modeling prediction result; and determining data verification results corresponding to the to-be-verified business data, the internal modeling prediction result and the external modeling prediction result according to a preset classification model. According to the technical scheme, through internal modeling and external modeling, sufficient modeling can be carried out on a business mechanism with a complex mechanism, and through fusion modeling processing of the preset classification model, the mistake possibly existing in the business data can be accurately identified.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a data verification method, device, equipment, storage medium and product. Background Art

[0002] In the financial audit scenario, there is a need to verify the business data reported by financial institutions. On the one hand, the business data reported by financial institutions may be misplaced, filled in incorrectly, etc. due to human errors and other reasons; on the other hand, the reported business data may be tampered with by humans. Therefore, how to accurately verify the business data reported by financial institutions has become a technical problem that needs to be solved urgently. Summary of the invention

[0003] The present invention provides a data verification method, device, equipment, storage medium and product, so as to verify the business data reported by the organization through an internal and external joint mechanism, and accurately identify possible errors in the business data.

[0004] According to one aspect of the present invention, a data verification method is provided, the method comprising:

[0005] Obtain the business data to be verified at the target time point of the target institution, as well as the target internal prediction model set and the target external prediction model set; wherein the target internal prediction model set is obtained based on the internal modeling of the first historical business data of the target institution; the target external prediction model set is obtained based on the external modeling of the first historical business data and the second historical business data of the same type of institution as the target institution;

[0006] Based on the business data to be verified and the verified data of similar institutions, the target time point is predicted according to the target internal prediction model set and the target external prediction model set, and the corresponding internal modeling prediction results and external modeling prediction results are obtained respectively;

[0007] The data verification results corresponding to the business data to be verified, the internal modeling prediction results and the external modeling prediction results are determined according to the preset classification model.

[0008] According to another aspect of the present invention, there is provided a data verification device, the device comprising:

[0009] An acquisition module is used to acquire the business data to be verified at a target time point of a target institution, as well as a target internal prediction model set and a target external prediction model set; wherein the target internal prediction model set is obtained based on an internal modeling of the first historical business data of the target institution; and the target external prediction model set is obtained based on an external modeling of the first historical business data and a second historical business data of a similar institution to which the target institution belongs;

[0010] The prediction module is used to predict the target time point based on the business data to be verified and the verification data of similar institutions, according to the target internal prediction model set and the target external prediction model set, and obtain the corresponding internal modeling prediction results and external modeling prediction results respectively;

[0011] The verification module is used to determine the data verification results corresponding to the business data to be verified, the internal modeling prediction results and the external modeling prediction results according to the preset classification model.

[0012] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0013] at least one processor; and

[0014] a memory communicatively connected to the at least one processor; wherein,

[0015] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the data verification method described in any embodiment of the present invention.

[0016] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the data verification method described in any embodiment of the present invention when executed.

[0017] According to another aspect of the present invention, a computer program product is provided. The computer program product includes a computer program. When the computer program is executed by a processor, the data verification method according to any embodiment of the present invention is implemented.

[0018] The technical solution of the embodiment of the present invention is obtained by obtaining the business data to be verified at the target time point of the target institution and the target internal prediction model set and the target external prediction model set; wherein, the target internal prediction model set is obtained based on the internal modeling of the first historical business data of the target institution; the target external prediction model set is obtained based on the external modeling of the first historical business data and the second historical business data of the same institution as the target institution; based on the business data to be verified and the verification data of each similar institution, the target time point is predicted according to the target internal prediction model set and the target external prediction model set, respectively, and the corresponding internal modeling prediction results and external modeling prediction results are obtained respectively; the data verification results corresponding to the business data to be verified, the internal modeling prediction results and the external modeling prediction results are determined according to the preset classification model. This technical solution can fully model the complex business mechanism of the institution through internal modeling and external modeling, learn the operating characteristics of the institution itself, and then accurately identify possible errors in the business data through the fusion modeling processing of the preset classification model, thereby assisting the regulatory agency to timely discover related risks.

[0019] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 is a flow chart of a data verification method provided according to Embodiment 1 of the present invention;

[0022] Figure 2 is a flow chart of a data verification method provided according to Embodiment 2 of the present invention;

[0023] Figure 3 is a structural schematic diagram of a data verification device provided according to Embodiment 3 of the present invention;

[0024] Figure 4 It is a structural schematic diagram of an electronic device for implementing the data verification method of an embodiment of the present invention. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention 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 interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] Embodiment 1

[0028] Figure 1 A flowchart of a data verification method is provided for the first embodiment of the present invention. This embodiment can be applied to verify the business data reported by the organization to identify the erroneous data therein. The method can be executed by a data verification device, which can be implemented in the form of hardware and / or software, and the data verification device can be configured in an electronic device. Figure 1 As shown, a data verification method provided in this embodiment includes the following steps:

[0029] S110. Obtain the business data to be verified at the target time point of the target institution, as well as the target internal prediction model set and the target external prediction model set; wherein, the target internal prediction model set is obtained based on the internal modeling of the first historical business data of the target institution; and the target external prediction model set is obtained based on the external modeling of the first historical business data and the second historical business data of the same type of institution as the target institution.

[0030] The target institution may refer to an institution to be verified, for example, a regulated banking financial institution. The target time point may refer to the time point corresponding to the target institution's reported data for the current period.

[0031] The business data to be verified can be understood as the business data reported by the target institution in this period, which may include but is not limited to: standardized data of regulatory data (Examination and Analysis System Technology, EAST), off-site regulatory reports (such as 1104 reports), etc. Furthermore, the business data to be verified may include several business fields (i.e. business dimensions), each of which corresponds to a data item (i.e. field value). For example, the 1104 report may include business fields such as investment in interbank certificates of deposit and dividends payable.

[0032] The target internal prediction model set includes several target internal prediction models, each target internal prediction model is associated with a corresponding first business field, and can be obtained by internal modeling based on the first historical business data of the target organization. The first historical business data may include historical business data for each historical time point reported by the target organization.

[0033] The target external prediction model set includes several target external prediction models, each of which is associated with a corresponding second business field, and can be externally modeled based on the first historical business data of the target organization and the second historical business data of similar organizations to which the target organization belongs. Among them, similar organizations refer to organizations that belong to the same category as the target organization; the second historical business data can include historical business data for each historical time point reported by each similar organization.

[0034] In an embodiment of the present invention, during the supervision process of the target organization, the business data to be verified for the target time point reported by the target organization can be obtained, and at the same time, a target internal prediction model set and a target external prediction model set obtained in advance based on internal modeling and external modeling respectively can be obtained, so that the target internal prediction model set and the target external prediction model set can be used to predict the business data at the target time point respectively.

[0035] It is important to understand that by using internal modeling, it is possible to discover erroneous data that only modifies a small amount of data and fails to achieve self-consistency in the organization's internal operating logic; by using external modeling, it is possible to discover possible errors in the data reported by the target organization through the business data and operating methods of similar institutions to which the target organization belongs.

[0036] S120. Based on the business data to be verified and the verified data of similar institutions, the target time point is predicted according to the target internal prediction model set and the target external prediction model set, and the corresponding internal modeling prediction results and external modeling prediction results are obtained respectively.

[0037] Among them, the verified data can be understood as the business data of similar institutions that have passed the verification for the target time point, that is, the verified data can be considered as the real business data of similar institutions, which can serve as an important reference for discovering erroneous data in the data reported by the target institution.

[0038] The internal modeling prediction result may include the prediction result of each target internal prediction model in the target internal prediction model set for the target time point. Similarly, the external modeling prediction result may include the prediction result of each target external prediction model in the target external prediction model set for the target time point.

[0039] In an embodiment of the present invention, based on the business data to be verified of the target institution, each target internal prediction model in the target internal prediction model set can be used to predict the business data at the target time point, respectively, to obtain the first target prediction results output by the model, and each first target prediction result can be used as the internal modeling prediction result corresponding to the target institution at the target time point. Similarly, based on the business data to be verified of the target institution and the verification data of the same type of institution as the target institution, each target external prediction model in the target external prediction model set can be used to predict the business data at the target time point, respectively, to obtain the second target prediction results output by the model, and each second target prediction result can be used as the external modeling prediction result corresponding to the target institution at the target time point.

[0040] Furthermore, in order to eliminate the magnitude differences between data items in different business fields in the business data to be verified and the verified data, before using the business data to be verified of the target organization and the verified data of similar organizations for prediction, a pre-configured standardization processing method can be called to perform standardization processing on the business data to be verified and the verified data respectively.

[0041] S130. Determine data verification results corresponding to the business data to be verified, the internal modeling prediction results, and the external modeling prediction results according to the preset classification model.

[0042] The preset classification model may refer to a model used to identify erroneous data, which may include but is not limited to: BERT (Bidirectional Encoder Representations from Transformers) model, CatBoost model, XGB (eXtreme Gradient Boosting) model, etc. The data verification result can be used to characterize whether the business data reported by the target institution in this period contains errors.

[0043] In an embodiment of the present invention, the acquired business data to be verified, the internal modeling prediction results and the external modeling prediction results can be jointly input into a pre-trained preset classification model, and the model can be used for fusion modeling processing to identify erroneous data that may exist in the business data to be verified, thereby obtaining corresponding data verification results; wherein, the data verification results output by the model can be represented by 0 or 1, with 0 indicating the presence of an error and 1 indicating the absence of an error.

[0044] It should be understood that this embodiment does not specifically limit the classification logic of the preset classification model. In one embodiment, the classification logic of the preset classification model can be to analyze and process the data items of a single business field in the business data to be verified separately, and as long as there is a data item corresponding to a business field that is identified as having an error, the data verification result output by the final model is that there is an error. In another embodiment, the classification logic of the preset classification model can also be to analyze and process the data items of all business fields in the business data to be verified as a whole. If the deviation of the actual overall situation of the data items corresponding to all business fields from the internal modeling prediction results and the external modeling prediction results is greater than the preset deviation threshold, the data verification result output by the final model is that there is an error.

[0045] The technical solution of the embodiment of the present invention is obtained by obtaining the business data to be verified at the target time point of the target institution and the target internal prediction model set and the target external prediction model set; wherein, the target internal prediction model set is obtained based on the internal modeling of the first historical business data of the target institution; the target external prediction model set is obtained based on the external modeling of the first historical business data and the second historical business data of the same institution as the target institution; based on the business data to be verified and the verification data of each similar institution, the target time point is predicted according to the target internal prediction model set and the target external prediction model set, respectively, and the corresponding internal modeling prediction results and external modeling prediction results are obtained respectively; the data verification results corresponding to the business data to be verified, the internal modeling prediction results and the external modeling prediction results are determined according to the preset classification model. This technical solution can fully model the complex business mechanism of the institution through internal modeling and external modeling, learn the operating characteristics of the institution itself, and then accurately identify possible errors in the business data through the fusion modeling processing of the preset classification model, thereby assisting the regulatory agency to timely discover related risks.

[0046] Furthermore, based on the above-mentioned embodiments of the invention, this embodiment provides a data verification method, which further includes:

[0047] Obtain the historical business data of each institution at each historical time point, and standardize the historical business data to obtain historical business samples;

[0048] Based on the business fields in the preset regulatory data table, historical business samples are used to cluster the institutions to determine the target category corresponding to each institution.

[0049] The historical business data may refer to the business data reported by each institution in history. Standardization processing may be used to eliminate the magnitude differences between data items in different business fields. Standardization processing methods may include but are not limited to range standardization method, Z-score standardization method, etc.

[0050] A historical business sample represents standardized business data corresponding to a historical time point. A historical business sample may include information on different business dimensions of an organization at a historical time point. For example, a historical business sample of XX month of XX year may include the following business data: {(business field 1, data item); (business field 2, data item); (business field 3, data item)}.

[0051] The preset regulatory data table may refer to the regulatory business report submitted by each institution, which may include but is not limited to 1104 report, EAST report, People's Bank of China centralized report, etc.; the preset regulatory data table contains different business fields, each business field represents a business type of different dimensions, and the business fields covered by each institution may be different, that is, not every business field, every institution has a corresponding business, if the institution has no related business, the data item of the corresponding business field is empty.

[0052] The target category can be understood as the category corresponding to each institution after the institutions are clustered, and each institution corresponds to a unique target category. At the same time, a group of institutions belonging to the same target category have the same business fields for reporting business data.

[0053] In an embodiment of the present invention, all institutions can be clustered based on the historical business samples of each regulated institution, so as to obtain the target category corresponding to each institution. Specifically, the historical business data reported by each regulated institution for each historical time point can be obtained from a data storage location such as a local or cloud server, and then the data items corresponding to each business field in the above historical business data are standardized to obtain the historical business samples of each institution corresponding to different historical time points; then, all institutions can be divided into several categories based on the business fields in the preset regulatory data table and combined with the business type of each institution, so as to determine the target category corresponding to each institution.

[0054] Further, on the basis of the above-mentioned embodiment of the invention, based on the business fields in the preset supervision data table, each institution is clustered using historical business samples to determine the target category corresponding to each institution, including:

[0055] Step 1: Roughly classify each institution according to whether or not they have the corresponding business fields;

[0056] Step 2: determine whether the number of institutions included in each first category after rough classification is greater than a preset threshold of the number of institutions;

[0057] Step 3: If yes, call the preset clustering algorithm to cluster the historical business samples of each institution in the first category, count the number of samples of each institution's historical business samples belonging to each second category after clustering, and take the second category with the largest number of samples as the target category of the corresponding institution;

[0058] Step 4: If not, the first category is used as the target category of the corresponding institution.

[0059] Among them, coarse classification can be understood as a rough classification of all institutions based on the presence or absence of corresponding business fields of each institution. After the coarse classification, the category corresponding to each institution is recorded as the first category; it can be understood that the number of first categories corresponding to the coarse classification is usually not too large.

[0060] The preset clustering algorithm may be a pre-configured clustering algorithm for further fine-classifying the institutions after the coarse classification. The preset clustering algorithm may include but is not limited to: a hierarchical clustering algorithm, a K-means clustering algorithm, etc.; at the same time, several categories obtained after processing by the preset clustering algorithm may be recorded as the second category.

[0061] In an embodiment of the present invention, the process of roughly classifying and clustering institutions is as follows: first, each institution is roughly classified according to the presence or absence of the corresponding business field of each institution, that is, a group of institutions containing the same business field are divided into one category, and the first category corresponding to each institution is obtained after rough classification; then, the number of institutions of all institutions in each first category is counted, and compared with a preset institution number threshold (for example, 2); if the number of institutions corresponding to a certain type of institution is greater than the preset institution number threshold, the preset clustering algorithm is called to cluster the institutions of this type to obtain the target category corresponding to each institution in this type of institution; conversely, if the number of institutions corresponding to a certain type of institution is less than or equal to the preset institution number threshold, there is no need to cluster this type of institution, and the first category corresponding to each institution in this type of institution can be directly used as the corresponding target category.

[0062] Among them, the specific process of calling a preset clustering algorithm to cluster the institutions of this type to obtain the target category corresponding to each institution in this type of institution may include: when the number of institutions corresponding to a certain type of institution is greater than the preset institution number threshold, a preset clustering algorithm such as hierarchical clustering may be called to cluster the historical business samples of this type of institution at each historical time point, and after clustering, several second categories may be obtained; then, the number of samples of the historical business samples of each institution belonging to each second category may be counted, that is, after all the historical business samples of a certain institution are divided into each second category, the number of historical business samples contained in each second category; finally, the corresponding second category with the largest number of samples may be used as the target category of the corresponding institution.

[0063] Further, based on the above-mentioned embodiment of the invention, the process of obtaining a target internal prediction model set based on internal modeling includes:

[0064] Step 1: Obtain the target category corresponding to the target organization, and extract data items corresponding to each business field of the target category from the first historical business data;

[0065] Step 2: Divide the data items corresponding to each business field into training data items and test data items;

[0066] Step 3: Select one of the business fields as the current field in turn, use the training data item corresponding to the current field as the model output, and use the training data items corresponding to other business fields as the model input to build an internal prediction model corresponding to the current field;

[0067] Step 4: Use the test data items corresponding to each business field to evaluate each internal prediction model, and obtain the first prediction accuracy of each internal prediction model;

[0068] Step 5: Using a first preset number of internal prediction models with the highest first prediction accuracy as target internal prediction models in the target internal prediction model set.

[0069] Among them, the internal prediction model may refer to a model constructed based on the first historical business data of the target organization for predicting the data item corresponding to a certain business field. The internal prediction model is associated with the corresponding business field, that is, for each business field contained in the corresponding target category of the target organization, an internal prediction model is created accordingly.

[0070] In the embodiment of the present invention, the internal modeling process of the target mechanism specifically includes:

[0071] (1) The target category obtained by clustering the target organization in advance can be obtained, and based on the target category, the data items corresponding to each business field can be extracted from the first historical business data obtained. Each business field corresponds to data items at multiple historical time points. For example, business field 1 corresponds to one data item in January 2024 and one data item in February 2024. Alternatively, it can also be understood that each historical time point corresponds to data items of multiple business fields, i.e., multiple data dimensions.

[0072] (2) Divide the data items corresponding to each business field into training data items and test data items according to the preset data item segmentation ratio. The training data items are used to train and build the internal prediction model, and the test data items are used to evaluate the prediction performance of the internal prediction model. The preset data item segmentation ratio can be set accordingly according to actual needs, for example, it can be 7:3, 8:2, etc., and this embodiment does not specifically limit this.

[0073] (3) For each business field corresponding to the target organization, select one of the business fields as the current field in turn, use the training data item corresponding to the current field as the target output of the internal prediction model to be constructed, use the training data items corresponding to the remaining business fields except the current field as the model input, and construct and train the internal prediction model corresponding to the current field. After executing this step, a corresponding internal prediction model will be established for each business field corresponding to the target organization. Among them, the internal prediction model can use a deep neural network, such as a transformer model, or a traditional machine learning algorithm, such as a linear regression algorithm, an XGB model, etc. This embodiment does not impose specific restrictions on this.

[0074] (4) After establishing the internal prediction model corresponding to each business field, the internal prediction model can be evaluated using the test data items corresponding to each business field to obtain the first prediction accuracy rate corresponding to each internal prediction model.

[0075] (5) Determine a first preset number of internal prediction models with the highest first prediction accuracy as target internal prediction models, and save the above target internal prediction models into a target internal prediction model set, so as to be used for internal prediction of data items corresponding to business fields when it is necessary to verify the business data reported by the target organization. It should be understood that the first preset number is less than or equal to the number of business fields corresponding to the target organization.

[0076] Further, based on the above-mentioned embodiment of the invention, the process of obtaining a target external prediction model set based on external modeling includes:

[0077] Step 1: Obtain a target category corresponding to the target organization, and extract data items corresponding to each business field of the target category from the first historical business data and the second historical business data respectively;

[0078] Step 2: Divide the data items corresponding to each business field into training data items and test data items;

[0079] Step 3: Select one of the business fields in turn as the current field, and based on the training data items of the current field of each similar institution in the second historical business data, call the preset cluster data feature calculation formula to determine the cluster data training feature corresponding to the current field, and use the training data item corresponding to the current field in the first historical business data as the model output, and the cluster data training feature as the model input, so as to construct an external prediction model corresponding to the current field;

[0080] Step 4: evaluate each external prediction model using the test data items corresponding to each business field and the cluster data test features to obtain the second prediction accuracy of each external prediction model;

[0081] Step 5: Use a second preset number of external prediction models with the highest second prediction accuracy as target external prediction models in the target external prediction model set.

[0082] The cluster data training feature can be understood as the result determined based on the preset cluster data feature calculation formula using the training data items of the corresponding business fields of the same type of institutions as the target institution for the business fields of the target institution. Correspondingly, the cluster data test feature represents the result determined based on the preset cluster data feature calculation formula using the test data items of the corresponding business fields of the same type of institutions as the target institution for the business fields of the target institution.

[0083] In the embodiment of the present invention, the external modeling process of the target mechanism specifically includes:

[0084] (1) A target category obtained by clustering the target organization in advance may be obtained, and based on the target category, data items corresponding to each business field may be extracted from the first historical business data and the second historical business data respectively.

[0085] (2) Divide the data items corresponding to each business field into training data items and test data items according to a preset data item segmentation ratio. The preset data item segmentation ratios used in the external modeling and internal modeling stages may be the same or different, and this embodiment does not limit this.

[0086] (3) For each business field corresponding to the target organization, one of the business fields is selected as the current field in turn, and then the test data items of the current field of the same type of organization as the target organization are used to call the preset cluster data feature calculation formula to determine the cluster data training feature corresponding to the current field, where the preset cluster data feature calculation formula is expressed as follows:

[0087]

[0088] In the formula, represents the number of institutions in the i-th category; m represents the m-th target institution; j represents the j-th business field (current field); The data item representing the business field j corresponding to the institution k of the same category i as the target institution m; represents the cluster data characteristics of the target organization m corresponding to the business field j for the i-th organization. Further, if is the training data item, then the formula is the cluster data training feature; similarly, if is the test data item, then the formula That is the cluster data test feature.

[0089] Next, the training data item corresponding to the current field in the first historical business data can be used as the target output of the external prediction model to be constructed, and the cluster data training features calculated for the current field can be used as the model input to construct and train the external prediction model corresponding to the current field. Similarly, after executing this step, a corresponding external prediction model will be established for each business field corresponding to the target organization. It should be understood that the external prediction model and the internal prediction model can adopt the same model construction method or different model construction methods, and this embodiment does not impose specific restrictions on this.

[0090] (4) Based on the method of obtaining the cluster data training features in the aforementioned steps, the corresponding cluster data test features can be determined using the test data items of each business field of the same organization, and then the test data items corresponding to each business field (for the first historical business data) and the cluster data test features can be used to perform model evaluation on the above-mentioned external prediction model, thereby obtaining the second prediction accuracy corresponding to each external prediction model.

[0091] (5) The second preset number of external prediction models with the highest second prediction accuracy are determined as target external prediction models, and each of the above target external prediction models is saved in the target external prediction model set, so as to be used for external prediction of data items of corresponding business fields when it is necessary to verify the business data reported by the target organization. It should be understood that the second preset number is less than or equal to the number of business fields corresponding to the target organization, and the second preset number can be the same as or different from the first preset number, and this embodiment does not limit this.

[0092] It is worth noting that the internal modeling process and the external modeling process introduced above can be executed once before verifying the business data reported by the organization each time, so that the final data verification result is the most accurate, but this method requires more computing resources. Under normal circumstances, the business operated by the organization will not change significantly in the short term, that is, the business fields corresponding to the organization will not usually change in the short term. Therefore, the internal modeling process and the external modeling process can be executed once every preset model training cycle or after receiving the business adjustment notification of the organization, so that there is no need to execute the internal modeling process and the external modeling process every time the data is verified, which can reduce the consumption of computing resources. In addition, the internal modeling process and the external modeling process can be executed in parallel, thereby improving the efficiency of internal modeling and external modeling.

[0093] Further, on the basis of the above-mentioned embodiment of the invention, in S120, based on the business data to be verified and the verification data of each similar institution, the target time point is predicted according to the target internal prediction model set and the target external prediction model set, and the corresponding internal modeling prediction results and external modeling prediction results are obtained respectively, including:

[0094] S1201, obtaining the verification data of each similar institution and the target category corresponding to the target institution, and extracting the data items corresponding to each business field of the target category from the business data to be verified and the verification data;

[0095] S1202, respectively extracting a first business field corresponding to each target internal prediction model from the target internal prediction model set, and extracting a second business field corresponding to each target external prediction model from the target external prediction model set;

[0096] S1203, sequentially selecting one of the first business fields as the first current field, and using the data items corresponding to the other business fields as the model input of the corresponding target internal prediction model, to obtain the first target prediction result output by the target internal prediction model;

[0097] S1204, sequentially selecting one of the second business fields as the second current field, based on the data items of the second current fields corresponding to the same type of institutions, calling the preset cluster data feature calculation formula to determine the cluster data feature corresponding to the second current field, and using the cluster data feature as the model input of the corresponding target external prediction model to obtain the second target prediction result output by the target external prediction model;

[0098] S1205. Taking each first target prediction result as an internal modeling prediction result, and taking each second target prediction result as an external modeling prediction result.

[0099] The first business field represents a business field associated with the target internal prediction model, and the second business field represents a business field associated with the target external prediction model.

[0100] In the embodiment of the present invention, the method for obtaining the internal modeling prediction result and the external modeling prediction result specifically includes:

[0101] (1) The target category obtained after clustering of the target organization in advance can be obtained, and the verification data reported by the same type of organization as the target organization can be obtained from a data storage location such as a local or cloud server, and the data items corresponding to each business field of the target category can be extracted from the business data to be verified and the verification data.

[0102] (2) Extracting the first business field associated with each target internal prediction model and the second business field associated with each target external prediction model from the target internal prediction model set and the target external prediction model set respectively.

[0103] (3) For each first business field, one of the business fields is selected in turn as the first current field, and the data items corresponding to the other business fields except the first current field are used as the model input of the corresponding target internal prediction model (for the first current field) to obtain the first target prediction result output by the target internal prediction model. After executing this step, the first target prediction result of the target internal prediction model corresponding to each first business field will be obtained.

[0104] (4) Similarly, for each second business field, one of the business fields is selected in turn as the second current field, and the data items corresponding to the second current field of each similar organization are used to call the preset cluster data feature calculation formula to determine the cluster data feature corresponding to the second current field, and then the cluster data feature is used as the model input of the corresponding target external prediction model (for the second current field) to obtain the second target prediction result output by the target external prediction model. Similarly, after executing this step, the second target prediction result of the target external prediction model corresponding to each second business field will be obtained.

[0105] (5) The first target prediction results obtained can be used together as the internal modeling prediction results, and the second target prediction results obtained can be used together as the external modeling prediction results.

[0106] Furthermore, the acquired business data to be verified, the internal modeling prediction results and the external modeling prediction results can be input into a pre-trained preset classification model to obtain data verification results for the business data to be verified.

[0107] Embodiment 2

[0108] Figure 2 This is a flow chart of a data verification method provided in Example 2 of the present invention. Based on the above examples, this example provides an implementation of a data verification method, which can model the complex business mechanism of financial institutions and discover unreasonable data performance therein, thereby accurately identifying erroneous data in the business data reported by financial institutions. Figure 2 As shown, a data verification method provided by Embodiment 2 of the present invention specifically includes the following steps:

[0109] S210. Obtain the historical supervision data table reported by each institution, standardize the historical business data in the historical supervision data table, and obtain the historical business samples corresponding to each historical time point.

[0110] The regulatory data table includes several business fields, each of which corresponds to a different business type, and each of which corresponds to a data item.

[0111] S220. Based on the presence or absence of business fields in the historical supervision data table, the institutions are clustered using historical business samples to determine the target category corresponding to each institution.

[0112] S230. Perform internal modeling based on the data items of each business field corresponding to the target institution in the historical supervision data table to obtain a target internal prediction model set corresponding to the target institution.

[0113] The internal modeling process of the target mechanism may refer to the above-mentioned embodiment and will not be described in detail here.

[0114] S240. Perform external modeling based on the data items of each business field of the target institution and similar institutions in the historical supervision data table to obtain a set of target external prediction models corresponding to the target institution.

[0115] The external modeling process of the target mechanism may refer to the above-mentioned embodiment and will not be described in detail here.

[0116] S250, obtaining the business data to be verified of the target organization at the target time point.

[0117] S260. Based on the business data to be verified and the verification data of similar institutions, the target time point is predicted using the target internal prediction model set and the target external prediction model set to obtain the corresponding internal modeling prediction results and external modeling prediction results.

[0118] S270. The acquired business data to be verified, the internal modeling prediction results and the external modeling prediction results are input into a pre-trained preset classification model to obtain a data verification result for the business data to be verified.

[0119] In an embodiment of the present invention, a preset classification model can be used to perform fusion modeling processing on the internal modeling prediction results and the external modeling prediction results, so as to identify erroneous data that may exist in the business data to be verified.

[0120] The technical solution of the embodiment of the present invention can fully model the complex business mechanisms of an organization through internal modeling and external modeling, and can learn the operating characteristics of the organization itself better than the previous rule-based approach, thereby revealing the untruthfulness of the data; and then through the fusion modeling processing of the preset classification model, it can accurately identify possible errors in the business data, thereby assisting regulatory agencies to promptly discover related risks.

[0121] Embodiment 3

[0122] Figure 3 This is a schematic diagram of the structure of a data verification device provided by Embodiment 3 of the present invention. Figure 3 As shown, the device comprises:

[0123] An acquisition module 31 is used to acquire the business data to be verified at a target time point of a target institution, as well as a target internal prediction model set and a target external prediction model set; wherein the target internal prediction model set is obtained based on the internal modeling of the first historical business data of the target institution; and the target external prediction model set is obtained based on the external modeling of the first historical business data and the second historical business data of the same type of institution as the target institution;

[0124] The prediction module 32 is used to predict the target time point based on the business data to be verified and the verification data of each similar institution according to the target internal prediction model set and the target external prediction model set, and obtain the corresponding internal modeling prediction results and external modeling prediction results respectively;

[0125] The verification module 33 is used to determine the data verification results corresponding to the business data to be verified, the internal modeling prediction results and the external modeling prediction results according to the preset classification model.

[0126] The technical solution of the embodiment of the present invention obtains the business data to be verified at the target time point of the target institution and the target internal prediction model set and the target external prediction model set through the acquisition module; wherein the target internal prediction model set is obtained based on the internal modeling of the first historical business data of the target institution; the target external prediction model set is obtained based on the external modeling of the first historical business data and the second historical business data of the same institution as the target institution; the prediction module predicts the target time point according to the target internal prediction model set and the target external prediction model set based on the business data to be verified and the verification data of each similar institution, and obtains the corresponding internal modeling prediction results and external modeling prediction results respectively; the verification module determines the data verification results corresponding to the business data to be verified, the internal modeling prediction results and the external modeling prediction results according to the preset classification model. This technical solution can fully model the complex business mechanism of the institution through internal modeling and external modeling, learn the operating characteristics of the institution itself, and then accurately identify possible errors in the business data through the fusion modeling processing of the preset classification model, thereby assisting the regulatory agency to timely discover related risks.

[0127] Furthermore, based on the above-mentioned embodiment of the invention, the data verification device further includes:

[0128] The historical business sample acquisition module is used to obtain the historical business data of each institution at each historical time point and perform standardization on the historical business data to obtain historical business samples;

[0129] The clustering module is used to cluster the institutions based on the business fields in the preset regulatory data table using historical business samples to determine the target category corresponding to each institution.

[0130] Further, based on the above-mentioned embodiment of the invention, the clustering module includes:

[0131] A rough classification unit is used to roughly classify each institution according to whether or not the corresponding business field of each institution exists;

[0132] An institution quantity determination unit, used to determine whether the number of institutions included in each first category after rough classification is greater than a preset institution quantity threshold;

[0133] A first target category determination unit is configured to, if yes, call a preset clustering algorithm to cluster the historical business samples of each institution in the first category, count the number of samples of the historical business samples of each institution belonging to each second category after clustering, and use the second category with the largest number of samples as the target category of the corresponding institution;

[0134] The second target category determination unit is used to, if not, use the first category as the target category of the corresponding organization.

[0135] Further, based on the above-mentioned embodiment of the invention, the process of obtaining a target internal prediction model set based on internal modeling includes:

[0136] Obtaining a target category corresponding to the target organization, and extracting data items of each business field corresponding to the target category from the first historical business data;

[0137] Divide the data items corresponding to each business field into training data items and test data items;

[0138] Select one of the business fields as the current field in turn, use the training data item corresponding to the current field as the model output, and use the training data items corresponding to other business fields as the model input, so as to construct an internal prediction model corresponding to the current field;

[0139] Using the test data items corresponding to each business field to evaluate each internal prediction model, respectively, to obtain a first prediction accuracy rate of each internal prediction model;

[0140] A first preset number of internal prediction models with the highest first prediction accuracy are used as target internal prediction models in the target internal prediction model set.

[0141] Further, based on the above-mentioned embodiment of the invention, the process of obtaining a target external prediction model set based on external modeling includes:

[0142] Obtaining a target category corresponding to the target organization, and extracting data items of each business field corresponding to the target category from the first historical business data and the second historical business data respectively;

[0143] Divide the data items corresponding to each business field into training data items and test data items;

[0144] Select one of the business fields in turn as the current field, and based on the training data items of the current field of each similar institution in the second historical business data, call the preset cluster data feature calculation formula to determine the cluster data training feature corresponding to the current field, and use the training data item corresponding to the current field in the first historical business data as the model output, and the cluster data training feature as the model input, so as to construct an external prediction model corresponding to the current field;

[0145] Each external prediction model is evaluated using the test data items corresponding to each business field and the cluster data test features to obtain the second prediction accuracy of each external prediction model;

[0146] A second preset number of external prediction models with the highest second prediction accuracy are used as target external prediction models in the target external prediction model set.

[0147] Further, based on the above-mentioned embodiment of the invention, the prediction module 32 includes:

[0148] A data item extraction unit is used to obtain the verification data of each similar institution and the target category corresponding to the target institution, and extract the data items corresponding to each business field of the target category from the business data to be verified and the verification data;

[0149] A first business field and a second business field extraction unit, used to extract the first business field corresponding to each target internal prediction model from the target internal prediction model set, and to extract the second business field corresponding to each target external prediction model from the target external prediction model set;

[0150] A first target prediction result determination unit is used to select one of the first business fields in turn as the first current field, and use the data items corresponding to the other business fields as the model input of the corresponding target internal prediction model to obtain the first target prediction result output by the target internal prediction model;

[0151] A second target prediction result determination unit is used to select one of the second business fields as the second current field in turn, and based on the data items of the second current fields corresponding to the same type of institutions, call a preset cluster data feature calculation formula to determine the cluster data feature corresponding to the second current field, and use the cluster data feature as a model input of the corresponding target external prediction model to obtain a second target prediction result output by the target external prediction model;

[0152] The internal modeling prediction result and external modeling prediction result determination unit is used to use each first target prediction result as an internal modeling prediction result and each second target prediction result as an external modeling prediction result.

[0153] The data verification device provided in the embodiment of the present invention can execute the data verification method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0154] Embodiment 4

[0155] Figure 4A schematic diagram of an electronic device 40 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0156] like Figure 4 As shown, the electronic device 40 includes at least one processor 41, and a memory connected to the at least one processor 41, such as a read-only memory (ROM) 42, a random access memory (RAM) 43, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 41 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 42 or the computer program loaded from the storage unit 48 to the random access memory (RAM) 43. In the RAM 43, various programs and data required for the operation of the electronic device 40 can also be stored. The processor 41, the ROM 42, and the RAM 43 are connected to each other through a bus 44. An input / output (I / O) interface 45 is also connected to the bus 44.

[0157] A number of components in the electronic device 40 are connected to the I / O interface 45, including: an input unit 46, such as a keyboard, a mouse, etc.; an output unit 47, such as various types of displays, speakers, etc.; a storage unit 48, such as a disk, an optical disk, etc.; and a communication unit 49, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 49 allows the electronic device 40 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0158] The processor 41 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 41 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 41 performs the various methods and processes described above, such as a data verification method.

[0159] In some embodiments, the data verification method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 48. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 40 via the ROM 42 and / or the communication unit 49. When the computer program is loaded into the RAM 43 and executed by the processor 41, one or more steps of the data verification method described above may be performed. Alternatively, in other embodiments, the processor 41 may be configured to perform the data verification method in any other appropriate manner (e.g., by means of firmware).

[0160] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0161] In some embodiments, the data verification method can be implemented as a computer program, which is invisibly included in a computer program product, and the computer program implements the data verification method of the present invention when executed by a processor. The computer program product can be understood as a software product that mainly implements its solution through a computer program. The computer program for implementing the method of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the computer program, when executed by the processor, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The computer program can be executed entirely on the machine, partially on the machine, partially on the machine as an independent software package, and partially on a remote machine, or completely on a remote machine or server.

[0162] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, device, or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0163] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0164] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0165] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0166] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0167] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A data verification method, characterized in that: The method comprises: Obtaining the business data to be verified at the target time point of the target institution, as well as the target internal prediction model set and the target external prediction model set; wherein the target internal prediction model set is obtained based on the internal modeling of the first historical business data of the target institution; and the target external prediction model set is obtained based on the external modeling of the first historical business data and the second historical business data of the same type of institution as the target institution; Based on the business data to be verified and the verification data of each of the similar institutions, the target time point is predicted according to the target internal prediction model set and the target external prediction model set, and the corresponding internal modeling prediction results and external modeling prediction results are obtained respectively; The data verification results corresponding to the business data to be verified, the internal modeling prediction results and the external modeling prediction results are determined according to the preset classification model.

2. The method according to claim 1, characterized in that Also includes: Obtaining historical business data corresponding to each historical time point of each institution, and performing standardization processing on the historical business data to obtain historical business samples; Based on the business fields in the preset supervision data table, the institutions are clustered using the historical business samples to determine the target category corresponding to each institution.

3. The method according to claim 2, characterized in that The clustering of the institutions using the historical business samples based on the business fields in the preset supervision data table to determine the target category corresponding to each institution includes: Roughly classify the institutions according to whether they have the corresponding business fields; Determining whether the number of institutions included in each first category after the rough classification is greater than a preset threshold of the number of institutions; If so, calling a preset clustering algorithm to cluster the historical business samples of each of the institutions in the first category, counting the number of samples of the historical business samples of each of the institutions belonging to each second category after clustering, and taking the second category with the largest number of samples as the target category of the corresponding institution; If not, the first category is used as the target category corresponding to the organization.

4. The method according to claim 1, characterized in that: The process of obtaining the target internal prediction model set based on the internal modeling includes: Acquire a target category corresponding to the target organization, and extract data items of each business field corresponding to the target category from the first historical business data; Dividing the data items corresponding to each of the business fields into training data items and test data items; Selecting one of the business fields as the current field in turn, using the training data item corresponding to the current field as the model output, and using the training data items corresponding to other business fields as the model input, so as to construct an internal prediction model corresponding to the current field; Using the test data items corresponding to the business fields to evaluate the internal prediction models, respectively, to obtain a first prediction accuracy rate of the internal prediction models; A first preset number of the internal prediction models with the highest first prediction accuracy are used as target internal prediction models in the target internal prediction model set.

5. The method according to claim 1, characterized in that The process of obtaining the target external prediction model set based on the external modeling includes: Acquire a target category corresponding to the target organization, and extract data items of each business field corresponding to the target category from the first historical business data and the second historical business data respectively; Dividing the data items corresponding to each of the business fields into training data items and test data items; Select one of the business fields in turn as the current field, and based on the training data items of the current field corresponding to each of the same type of institutions in the second historical business data, call a preset cluster data feature calculation formula to determine the cluster data training features corresponding to the current field, and use the training data items corresponding to the current field in the first historical business data as model outputs, and the cluster data training features as model inputs, so as to construct an external prediction model corresponding to the current field; Using the test data items corresponding to the business fields and the cluster data test features respectively to evaluate the external prediction models, and obtain the second prediction accuracy of the external prediction models; A second preset number of the external prediction models with the highest second prediction accuracy are used as target external prediction models in the target external prediction model set.

6. The method according to claim 1, characterized in that Based on the business data to be verified and the verification data of each of the similar institutions, the target time point is predicted according to the target internal prediction model set and the target external prediction model set, and the corresponding internal modeling prediction results and external modeling prediction results are obtained respectively, including: Obtaining the verification data of each of the same type of institutions and the target category corresponding to the target institution, and extracting data items of each business field corresponding to the target category from the business data to be verified and the verification data; Extracting the first business field corresponding to each target internal prediction model from the target internal prediction model set, and extracting the second business field corresponding to each target external prediction model from the target external prediction model set; Selecting one of the first business fields in turn as the first current field, and using data items corresponding to other business fields as model inputs of the corresponding target internal prediction model to obtain a first target prediction result output by the target internal prediction model; Selecting one of the second business fields in turn as the second current field, based on the data items of the second current field corresponding to each of the same type of institutions, calling a preset cluster data feature calculation formula to determine the cluster data feature corresponding to the second current field, and using the cluster data feature as the model input of the corresponding target external prediction model to obtain a second target prediction result output by the target external prediction model; Each of the first target prediction results is used as the internal modeling prediction result, and each of the second target prediction results is used as the external modeling prediction result.

7. A data verification device, characterized in that: The device comprises: An acquisition module is used to acquire the business data to be verified at a target time point of a target institution, as well as a target internal prediction model set and a target external prediction model set; wherein the target internal prediction model set is obtained based on an internal modeling of the first historical business data of the target institution; and the target external prediction model set is obtained based on an external modeling of the first historical business data and a second historical business data of a similar institution to which the target institution belongs; A prediction module, used to predict the target time point based on the business data to be verified and the verification data of each of the similar institutions according to the target internal prediction model set and the target external prediction model set, and obtain corresponding internal modeling prediction results and external modeling prediction results respectively; A verification module is used to determine the data verification results corresponding to the business data to be verified, the internal modeling prediction results and the external modeling prediction results according to a preset classification model.

8. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the data verification method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the data verification method according to any one of claims 1 to 6 when executed.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the computer program implements the data verification method according to any one of claims 1 to 6.

Images