RU2724799C1 - Information processing method for filling data model library and device for its implementation - Google Patents

Abstract

FIELD: computer equipment.SUBSTANCE: invention relates to computer engineering and particularly to a method and an apparatus for processing information for filling a model library of data models. Said technical result is achieved by implementing a method of processing information for filling a library of data models, performed by at least one computing device, comprising steps, on which information is obtained to form at least one data model, based on the obtained information, determining a type, a simulation method and a data model structure, determining sampling data to form a data model, based on the data sample, determining metrics to form a data model, generating a data model based on information on the type, simulation method, structure of the data model and metrics determined at the previous stage, validating at least one formed data model, storing the data model in the library of data models.EFFECT: high efficiency of forming models by increasing their accuracy.6 cl, 3 dwg

Description

FIELD OF TECHNOLOGY

[001] This technical solution generally relates to the field of computing, and in particular to a method and apparatus for processing information to populate a library of data models.

BACKGROUND

[002] Currently, in existing systems designed for processing data for model development, there are drawbacks related to the quality of the developed models, as a result of which the accuracy of the generated models and the speed of their development are reduced, and the resulting models do not have the required optimization parameters. The closest solution to the claimed solution is a system that allows you to develop and execute forecast models, disclosed in the application US 2017249556 (A1), publ. 08/31/2017. The known system includes: a data management server, which can be accessed through a web browser; data extraction tool; model validation tool for model validation by evaluating the data set of the analytical environment and the data set of the production environment; and a model execution tool that allows the user to choose when and how often the model will be validated.

[003] A disadvantage of the known solution is the low accuracy of the generated models and low quality indicators for their optimization. Also, the well-known solution does not have a set of functional properties that allow a higher-level product, for example, Sberbank Data Science, to carry out orchestration of the life cycle of models developed by the company from the "Initiative" stage to the stage of implementation of the model in commercial operation and continuous periodic monitoring of model quality on the corresponding data samples.

ESSENCE OF TECHNICAL SOLUTION

[004] A technical problem or task posed in this technical solution is to create a simple and reliable method and device for processing information to populate a library of data models. Also, the presented solution provides the orchestration of a single model life cycle from initiating an idea to introducing a model into an industrial execution environment and periodically monitoring the quality of models.

[005] The technical result achieved by solving the above technical problem is to increase the efficiency of forming models by increasing their accuracy.

[006] The specified technical result is achieved due to the implementation of the information processing method for filling the library of data models performed by at least one computing device, comprising the steps of:

- receive information for the formation of at least one data model;

- determine, based on the information received, the type, modeling method and structure of the data model;

- determine the selection of data to form a data model;

- based on a data sample, metrics are determined to form a data model;

- form a data model taking into account information about the type, modeling method, structure of the data model and the metrics defined at the previous stage;

- validate at least one generated data model;

- save the data model to the library of data models.

[007] In one particular embodiment of the method, the step of storing the data model in the data model library includes the steps of:

- form a block for storing the attributes of the model in the said library and store information on the types of the model and the results of its validation in this block;

- form a block for storing parameters in the said library and store information on the parameters of the data model that entities may have in this block;

- form a block for storing data about users and store information about users in this block;

- form an entity storage unit and create an entity instance for each attribute in the entity storage unit;

- form a block for storing entity parameters and save identifiers of identifiers, parameter identifiers, and parameter values for the entity into this block;

- form a block for storing historical data of entities and save identifiers of entities, identifiers of parameters and the history of changes in parameters.

[008] In another particular embodiment of the method, data model output is further obtained; comparing the obtained output data with the stored output data and determining that the data model is changed if at least one value of the obtained output deviates from the value of the stored output data.

[009] In another particular embodiment of the method, the life cycle of at least one data model is additionally monitored and the time spent by the data model at each stage of its formation is predicted.

[0010] In another particular embodiment of the method, the generated data model is additionally signed with at least one electronic signature.

[0011] In another preferred embodiment of the claimed solution, there is provided an information processing system for populating a library of data models containing at least one computing device and at least one memory containing machine-readable instructions that, when executed by at least one computing device, perform the above way.

BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The features and advantages of the present invention will become apparent from the following detailed description of the invention and the accompanying drawings, in which:

In FIG. 1 - presents an example implementation of an information processing system.

In FIG. 2 - presents an example implementation of a library of data models.

In FIG. 3 is an example of a general view of a computing device.

DETAILED DESCRIPTION OF THE INVENTION

[0013] Below will be described the concepts and terms necessary for understanding this technical solution.

[0014] In this technical solution, a system is meant, including a computer system, a computer (electronic computer), CNC (numerical control), PLC (programmable logic controller), computerized control systems, and any other devices that can perform a given , clearly defined sequence of operations (actions, instructions).

[0015] An instruction processing device is understood to mean an electronic unit or an integrated circuit (microprocessor) executing machine instructions (programs).

[0016] The command processing device reads and executes machine instructions (programs) from one or more data storage devices. Hard disk drives (HDD), flash memory, ROM (read-only memory), solid-state drives (SSD), optical drives can act as storage devices.

[0017] A program is a sequence of instructions for execution by a computer control device or an instruction processing device.

[0018] In accordance with the circuit of FIG. 1, the information processing system comprises interconnected information processing device 10 for filling a library of data models, a data storage device 20, and a data model library 30.

[0019] The information processing device 10 for filling the library of data models 30 may be implemented on the basis of at least one computing device and may include interconnected: data collection module 11; primary data processing module 12; module 13 determine the selection of data and metrics; a data model generation module 14; module 15 validation of the data model. The presented modules can be performed on the basis of the software and hardware of the information processing device 10, made in the software part so as to perform the functions assigned to them in this application.

[0020] At the first stage of operation of the information processing device 10, information is received to the data collection module 11 for generating at least one data model, which can be supplied to the said module 11 from at least one data source 1. The data source 1 can be at least one database (DB) in which information for generating at least one data model is contained, or the data source 1 can be a user device, for example, a portable or stationary computer, mobile phone or smartphone, tablet etc., connected to the data acquisition module 11. For example, by means of a user device, a request may be sent to the information processing device 10, according to which the user will be presented with an interface for entering information about the data model, which will subsequently be generated at the output of the said device 10.

[0021] In the presented interface, the user, depending on the access parameters, can also look at the model cards that are stored in the data storage device 20 and which contain information on the statuses of the models, in particular, whether they are under development, validation or operation , about their attributes, metrics, sets of entities and the life cycle of the model. The methods for generating attributes, metrics, and a set of entities of data models are widely known in the art and will not be described in more detail in this application (see, for example, Sergey A. Orlov, “Software Engineering. A Textbook for High Schools. 5th edition updated and supplemented. The Third Generation Standard ”, St. Petersburg: Peter, 2016, 640 pp.).

[0022] Also, the user can send a request to generate a new data model. To direct this request, the user can create a new model card through the interface for entering information, in which the user indicates, in particular, the model developer and the type of data model. Additionally, the user can specify the initiator, in particular the identifier of the customer’s unit or the contact person of the customer. Information about the type of data model characterizes the scope of its application, for example, indicates that the data model will be used to determine the risks of credit debt, etc. Additionally, the user can select specific attributes of the model and entity to form a set of entities, as well as specify life cycle parameters models i.e. choose a period (set time parameters) for which the data model formed in the future will be relevant.

[0023] Accordingly, after entering information for generating at least one data model, i.e. after filling out the model card and entering the appropriate command, the information is sent to the information processing device 10 and enters the data acquisition module 11. Module 11 stores the received information in the data storage device 20, in which all the cards of the models are stored, after which any user of the user device with the necessary access rights can familiarize themselves with the information of this card. Information on the user who sent the request, for example, its identifier, as well as information on the status of the data model, in particular information that the model is under development, can also be entered in the model card.

[0024] In an alternative implementation of the claimed solution, said information for generating at least one data model is extracted by the module 11 from the database of the data source 1 according to the program algorithm embedded therein and automatically generates the said model card in the data storage device 20.

[0025] Also, the data collection module 11 directs the obtained information to form the data model to the primary data processing module 12 of the model, which, based on the data of the model developer and the type of data model, determines the model code, as well as the modeling method and structure of the data model. To determine the modeling method and structure of the data model, said module 12 extracts data on the type of data model specified by the user from the received information, and identifies the model type, modeling method, and structure of the data model for the data type and code of the data model. Data on the types of data models, their types of codes and the corresponding identifiers of the model type, modeling method and structure of the data model can be predefined in the memory of said module 12 by at least one user of this module 12 by means of a specialized computing device 2.

[0026] If there is no information on the identifiers of the type of model, the modeling method, and the structure of the data model for the received data on the type of data model specified by the user in the memory of the module primary processing module 12, then module 12 may send a corresponding request to computing device 2, after which the user of the computing device 2 will be able to find the model card and enter information into the memory of the module 12 about the identifiers of the model type, modeling method and structure of the data model according to the type of data model specified by the user.

[0027] After determining the model type, modeling method, data model structure, the primary model data processing module 12 sends identifiers of the model type, modeling method, and data model structure to the data sample and metric determination module 13, which determines the data sample to form the data model. To determine the data sample, the module 13 sends to the computing device 2 data about the identifiers of the model type, modeling method and structure of the data model, in response to which the user of the computing device 2 sends the identifiers of the data sample for this type of model, modeling method and structure of the data model according to the established goals defined business objective and company strategy. Upon receipt of the said sample identifier, module 13 refers to the database with which it can be equipped to determine data metrics in accordance with the obtained identifier. Information about the identifiers of the samples and the corresponding data metrics can be pre-recorded in the database of module 13 by the user of the device 10. For more details, methods for determining the samples and metrics for the data model are disclosed, for example, in the article “Data Structures with Program Properties” located on the Internet at : https://habr.com/en/post/347856/.

[0028] The obtained metrics, identifiers of the model type, modeling method, and structure of the data model are then sent by module 13 to the data model generation module 14, which, based on the received data, is used to generate the data model based on methods well known in the art. Also, module 14 may be configured to adjust information about the status of the data model in said model card. For example, module 14 may indicate a status — the model is developed.

[0029] The generated data model is transmitted by said module 14 to the data model validation module 15, which directs the generated model to the data validation device 3. The user of device 3 checks metrics, effective indicators of the data model and other data of the model for compliance with the specified requirements of the business task and company strategy and sends to module 15 data on the successful completion of validation or that the data model has not passed validation. Said module 15 may also be adapted to adjust the status of the data model in said model card. For example, module 15 may set a status indicating that the data model is validated or has passed or failed validation. In an alternative implementation of the claimed solution, validation may be carried out by the methods disclosed in the closest analogue of US2017249556 (A1).

[0030] In the case of successful validation, the generated data model by said module 15 is sent to the model library 30 for storage, and the module 15 enters a link to the generated data model into the data model card stored in the data storage device 20. If the model does not pass the validation, then the module 15 sends a notification to the computing device 2 about the need to adjust the data model in that part that does not meet the specified requirements. In accordance with the notification received, the user of the device 2 can make changes to the generated model and send it again for validation.

[0031] Thus, due to the fact that the data model is formed taking into account information about the type, modeling method, structure of the data model and metrics, the specified technical result is achieved, which consists in increasing the efficiency of the formation of models by increasing their accuracy.

[0032] Accordingly, the generated data model according to the circuit shown in FIG. 2, enters the data processing module 31, which can also be implemented on the basis of at least one computing device and which extracts information about the model type and the results of its validation from the obtained data model and generates it in the said library, in particular, in the data storage module 32 block 33 storing the attributes of the model, which stores information about the types of the model and the results of its validation.

[0033] Next, the data processing module 31 likewise extracts from the generated data model information about: parameters of the data model that entities may have; users attributes entity identifiers, parameter identifiers, and parameter values for the entity. Accordingly, said module 31 forms in module 32:

- unit 34 for storing parameters, in which module 31 stores information about the parameters of the data model, which may have entities;

- a user data storage unit 35 to which user information is stored by the module 31;

an entity storage unit 36, in which module 31 creates an entity instance for each attribute;

- unit 37 for storing entity parameters into which module 31 stores entity identifiers, parameter identifiers, and parameter values for the entity

[0034] Also, the module 31 forms, in the data storage module 32, an entity historical data storage unit 38, in which the module 31 also generates a new record with information about entity identifiers, parameter identifiers and their values. Additionally, the identifier of the user who defined this parameter earlier and the date the record was created can be added to the new record. After that, the data processing module 31 accesses the data model card stored in the data storage device 20 and adjusts the status of the model to “developed” and the data model can be put into operation using methods known from the prior art. In addition, the data model generated in the library can be signed with at least one electronic signature, for example, the signature of a user of device 1 or a user of library 30 of data models.

[0035] Accordingly, upon receipt of a request to operate a data model, module 31 corrects the status of the model in the data storage device 20 to “model in use”. The operation of a data model means its transfer by methods well known in the art to an automated system that further collects and processes data according to the obtained data model.

[0036] During the operation of the data model, it may be necessary to adjust its parameters. To adjust the parameters, a corresponding request is sent to the data processing module 31 by the system user, containing the user identifier by which the request was initiated, at least one entity identifier, at least one parameter identifier of this entity, and a new value for this parameter. The aforementioned request to the module 31 can be sent by the administrator of the model library 30 or any other user of the user device with the appropriate access rights.

[0037] The data processing unit 31, based on the data on the entity identifier and the parameter identifier, finds the value of this parameter in the entity parameter storage unit 37 and changes it to a new parameter value. Also, the module 31 forms in the block 38 of the storage of historical data of entities a new record in which the identity of the entity, the identifier of the parameter and the history of the change of the parameter, in particular the previous parameter value and the new parameter value, are stored. Additionally, the identifier of the user who initiated the parameter change and the date of creation of the new record can be included in the record. Thus, in order to adjust the model parameters, the data processing module 31 does not need to analyze all the data in the model library 30, but rather turn to the entity parameter storage unit 37 in which these parameters are stored, thereby increasing the speed of data model corrections.

[0038] Additionally, data model output obtained as a result of its validation from data model validation module 15 may also be stored in data storage device 20, and data processing module 31 may be configured to detect changes to the data model in use. To this end, the data processing module 31 requests the model output from the automated system and compares the received output with the stored output. If the data processing unit 31 has determined that at least one value of the received output deviates from the value of the stored output, then the data model is considered changed. The information that the data model is changed, the module 31 sends to the data storage device 20 to display it in the data model card.

[0039] Also additionally, the data processing module 31 may be configured to track the life cycle of the model. For this, module 31 is configured to access publicly available data that contains information about the parameters and their current values. Module 31 records the change in the actual parameter values, compares them with the values stored in the model library 30, and for data models with different parameter values, module 31 sets the status of the model to the data model stored in the data storage device 20 “model is outdated”.

[0040] In addition, the information processing device 10 may be configured to predict the time spent by the data model at each stage of its formation. For this, the primary data processing module 12 can be equipped with a specialized database, which stores information about the average time of formation of the data model at each stage of its formation, depending on its type, modeling method and structure. Accordingly, after module 12 has determined the type of model, modeling method, structure of the data model, it turns to the database to determine the time of its formation at each stage based on the average time of formation of the data model. The obtained time for finding the data model at each stage of its formation is sent by the module 12 to the data storage device 20 for display in the data model card.

[0041] Thus, the presented solution provides a unification of the business processes of model management for company divisions, as a result of which the manageability and quality of models are improved, the timeliness of decision-making based on regular activities and the operation of the system is reduced, the costs associated with control and validation functions are reduced, as well as model risk.

[0042] In a general view (see FIG. 3), the computing device comprises one or more processors (201) connected by a common data bus, memory means such as RAM (202) and ROM (203), input / output interfaces (204) ), input / output devices (205), and a device for network communication (206).

[0043] A processor (201) (or multiple processors, a multi-core processor, and the like) may be selected from a variety of devices that are currently widely used, for example, manufacturers such as Intel ™, AMD ™, Apple ™, Samsung Exynos ™, MediaTEK ™, Qualcomm Snapdragon ™, etc. Under the processor or one of the processors used in the device (200), it is also necessary to take into account a graphic processor, for example, an NVIDIA or Graphcore GPU, the type of which is also suitable for complete or partial execution of the method, and can also be used for training and application of machine learning models in various information systems.

[0044] RAM (202) is a random access memory and is intended to store machine-readable instructions executed by the processor (201) to perform the necessary operations for logical data processing. RAM (202), as a rule, contains executable instructions of the operating system and corresponding software components (applications, program modules, etc.). At the same time, the available memory capacity of the graphics card or graphics processor can act as RAM (202).

[0045] The ROM (203) is one or more permanent storage devices, for example, a hard disk drive (HDD), a solid state drive (SSD), flash memory (EEPROM, NAND, etc.), optical storage media ( CD-R / RW, DVD-R / RW, BlueRay Disc, MD), etc.

[0046] Various types of I / O interfaces (204) are used to organize the operation of the components of the device (200) and organize the operation of external connected devices. The choice of appropriate interfaces depends on the particular computing device, which can be, but not limited to: PCI, AGP, PS / 2, IrDa, FireWire, LPT, COM, SATA, IDE, Lightning, USB (2.0, 3.0, 3.1, micro, mini, type C), TRS / Audio jack (2.5, 3.5, 6.35), HDMI, DVI, VGA, Display Port, RJ45, RS232, etc.

[0047] Various means (205) of I / O information, for example, a keyboard, a display (monitor), a touch screen, a touch pad, a joystick, a mouse, a light pen, a stylus, a touch, are used to provide user interaction with the device (200). panel, trackball, speakers, microphone, augmented reality, optical sensors, tablet, light indicators, projector, camera, biometric identification tools (retina scanner, fingerprint scanner, voice recognition module), etc.

[0048] The network interaction tool (206) provides data transmission via an internal or external computer network, for example, an Intranet, the Internet, a LAN, etc. As one or more means (206), but not limited to: an Ethernet card, GSM modem, GPRS modem, LTE modem, 5G modem, satellite communications module, NFC module, Bluetooth and / or BLE module, Wi-Fi module, etc.

[0049] Additionally, satellite navigation systems as part of the system (200), for example, GPS, GLONASS, BeiDou, Galileo, can also be used. The specific choice of elements of the device (200) for the implementation of various software and hardware architectural solutions may vary while maintaining the required functionality.

[0050] Modifications and improvements to the above-described embodiments of the present technical solution will be apparent to those skilled in the art. The preceding description is provided as an example only and is not subject to any restrictions. Thus, the scope of the present technical solution is limited only by the scope of the attached claims.

Claims (24)

1. A method of processing information to populate a library of data models, performed by at least one computing device, comprising stages in which - receive information for the formation of at least one data model; - determine, based on the information received, the type, modeling method and structure of the data model; - determine the selection of data to form a data model; - based on a data sample, metrics are determined to form a data model; - form a data model taking into account information about the type, modeling method, structure of the data model and the metrics defined at the previous stage; - validate at least one generated data model; - save the data model to the library of data models. 2. The method according to p. 1, characterized in that the step of saving the data model in the library of data models includes the steps at which - form a block for storing the attributes of the model in the said library and store information on the types of the model and the results of its validation in this block; - form a block for storing parameters in the said library and store information on the parameters of the data model that entities may have in this block; - form a block for storing data about users and store information about users in this block; - form an entity storage unit and create an entity instance for each attribute in the entity storage unit; - form a block for storing entity parameters and save identifiers of identifiers, parameter identifiers, and parameter values for the entity into this block; - form a block for storing historical data of entities and save identifiers of entities, identifiers of parameters and the history of changes in parameters. 3. The method according to p. 1, characterized in that it contains additional steps, in which - get the output of the data model; - comparing the received output with the stored output; - determine that the data model is changed in case of deviation of at least one value of the received output from the value of the stored output. 4. The method according to p. 1, characterized in that it contains additional steps, in which - track the life cycle of at least one data model; - predict the time spent by the data model at each stage of its formation. 5. The method according to p. 1, characterized in that it contains an additional step, which sign the generated data model with at least one electronic signature. 6. An information processing system for filling a library of a model of data models containing at least one computing device and at least one memory containing machine-readable instructions that, when executed by at least one computing device, perform the method according to any one of claims. 1-5.

Images