KR20250059146A - Method and Device for migrating data using Artificial Intelligence - Google Patents

Abstract

In a preferred embodiment of the present invention, a method of transferring data stored in a relational database management system (RDBMS) to a graph database management system (GraphDBMS) is proposed.

Description

AI-based data transfer device and method {Method and Device for migrating data using Artificial Intelligence}

The present invention relates to a method for smoothly transferring data from an RDBMS to a GraphDBMS through machine learning.

A database is a computerized information storage and retrieval system. A relational database management system (RDBMS) is a database management system (DBMS) that uses relational techniques to store and retrieve data.

US10169409B2

In a preferred embodiment of the present invention, data stored in a relational database management system (RDBMS) is to be transferred to a graph database management system (GraphDBMS).

As a preferred embodiment of the present invention, an AI-based data transfer device includes: a preprocessing unit that analyzes a schema of an RDBMS to identify tables, records constituting the tables, and foreign keys indicating relationships between records of each of a plurality of tables; a machine learning unit that generates data characteristics between records constituting each of the plurality of tables for data not explicitly defined in the schema of the RDBMS and infers relationships between the plurality of tables based on the generated data characteristics; a first mapping unit that maps the tables of the RDBMS identified in the preprocessing unit to labels of the GraphDBMS, the records of the RDBMS to nodes of the GraphDBMS, and the foreign keys of the RDBMS to relationships of the GraphDBMS, respectively; It is characterized by including a second mapping unit that automatically maps the table of the RDBMS inferred in the machine learning unit to the label of the GraphDBMS, the record of the RDBMS to the node of the GraphDBMS, and the foreign key of the RDBMS to the relation of the GraphDBMS; and a data conversion unit that transfers data of the RDBMS to the GraphDBMS according to the mapping of the first mapping unit and the second mapping unit.

As a preferred embodiment of the present invention, the AI-based data transfer device is characterized by further including a data verification unit that identifies and corrects data inconsistencies or omissions when transferring data from the data conversion unit.

As a preferred embodiment of the present invention, the data conversion unit is characterized by further including a translation unit that translates an SQL query of an RDBMS into a query language of a GraphDBMS or a Cypher query language.

In a preferred embodiment of the present invention, when transferring data stored in a relational database management system (RDBMS) to a graph database management system (GraphDBMS), data patterns are automatically identified through machine learning and data conversion is performed, thereby significantly reducing the resources required for data transfer. In addition, the accuracy of data transfer can be improved by automatically supplementing the integrity occurring during the data transfer process through an AI algorithm.

Figure 1 illustrates an example of a table of RDMBS.
FIG. 2 illustrates an internal configuration diagram of an AI-based data transfer device (200) as a preferred embodiment of the present invention.
FIG. 3 illustrates an example of a preferred embodiment of the present invention in which data features are extracted from the execution result data of an application through machine learning in an AI-based data transfer device (200) and then converted into GraphDBMS.
FIG. 4 illustrates an example of mapping data of RDBMS illustrated in FIG. 1 to be automatically converted into GraphDBMS, as a preferred embodiment of the present invention.

Hereinafter, various embodiments of the present invention will be described in more detail with reference to the attached drawings. Furthermore, the suffixes “part” and “device” for components used in the following description are simply given in consideration of the ease of writing this specification, and the “part” and “device” may be used interchangeably with each other, and may be designed as hardware or software depending on the embodiment. Furthermore, embodiments of the present invention will be described in detail with reference to the attached drawings and the contents described in the attached drawings, but the present invention is not limited or restricted by the embodiments.

Figure 1 illustrates an example of a table of RDMBS.

A relational database management system (RDBMS) consists of tables consisting of rows and columns of data. In an RDBMS, data and relationships are stored and managed using tables (110, 120, 130), records within the tables, and foreign keys (S131a~S133a, S131b~S133b) that represent relationships between records in each of multiple tables (110, 120, 130). A table schema represents a data structure created according to a table definition. For example, assume a table schema of R(A1, A2,...,An). R represents the name of the table, and A1, A2,...,An each represent the names of fields.

Referring to Figure 1, data in the RDBMS is managed in the form of tables (110, 120, 130). The first table (110), the second table (120), and the third table (130) each consist of 9 records.

The record (S111) of the first table (110) includes a primary key (PK) field (111) that identifies an employee and an employee name (112) field. The data of the primary key field that identifies an employee are 101 to 109. The data of the employee name field (112) are John, Jane, Mark, Emily, David, Sarah, Laura, Chris, and Jessica.

Each record constituting the second table (120) includes a primary key field (121) that identifies a department and a department name field (122) data. The data of the primary key field (121) that identifies a department is 801 to 809. The data of the department name field (122) is HR, Tech, R&D, Marketing, Finance, QA, Customer Care, Sales, and Operation.

Each record constituting the third table (130) includes data representing a relationship between an employee and a department in the company. The third table (130) stores data on an employee who has moved between departments in the company. The third table (130) is composed of a first reference key (Foreign Key, FK) (135) referencing the unique identification number key of the employee and a second reference key (FK) (136) referencing the unique identification number key of the department. The first reference key (135) must have a value in the primary key field (111) that identifies the employee included in the first table. The second reference key (136) must have a value in the primary key field (121) that identifies the employee included in the second table.

The third table (130) identifies rows of fields in one table among fields in another table by using a foreign key (FK). For example, if the data of the primary key field that identifies an employee is 104, it identifies "104 Emily" in the first table (110). Then, it identifies rows in the table corresponding to data 802, 803, and 806 (S131b, S132b, S133b) of other fields in the records (S131a, S132a, S133a) whose primary key field data is 104. For example, it identifies "802 Tech", "803 R&D", and "806 QA". Employee Emily can check her history of working in the company's Tech, R&D, and QA departments.

In another preferred embodiment of the present invention, the third table (130) may further include additional data, not shown in FIG. 1, such as all departments in which Emily has worked and the length of time she worked in each department.

FIG. 2 illustrates an internal configuration diagram of an AI-based data transfer device (200) as a preferred embodiment of the present invention.

The AI-based data transfer device (200) includes a preprocessing unit (210), a machine learning unit (220), a mapping unit (230), and a data conversion unit (240). In addition, the AI-based data transfer device (200) may further include a data verification unit (250).

The preprocessing unit (210) analyzes the table schema of the RDMBS as in the example of Fig. 1 to identify tables, records constituting the tables, and foreign keys indicating relationships between records of each of the multiple tables.

The machine learning unit (220) generates data features between records constituting each of the plurality of tables for data not explicitly defined in the schema of the RDBMS, and infers relationships between the plurality of tables based on the generated data features.

The machine learning unit (220) can also generate data features between records constituting each of the plurality of tables based on data identified from the execution results of the data analysis application, and infer relationships between the plurality of tables based on the generated data features. One example of this is shown in FIG. 3.

The mapping unit (230) maps tables of the RDBMS to labels of the GraphDBMS, records of the RDBMS to nodes of the GraphDBMS, and foreign keys of the RDBMS to relationships of the GraphDBMS.

As another preferred embodiment of the present invention, the mapping unit (230) may include a first mapping unit (232) and a second mapping unit (234).

The first mapping unit (232) performs mapping based on the results received from the preprocessing unit (210), and the second mapping unit (234) performs mapping based on the results received from the machine learning unit (220).

An example of receiving an SQL log of an RDBMS like Fig. 1 in the first mapping section (232) and outputting it as a Cypher log of a GraphDBMS like Fig. 4 is explained as follows.

The first mapping unit (232) maps the table of the RDBMS identified based on the table schema in the preprocessing unit (210) to the label of the GraphDBMS, the record of the RDBMS to the node of the GraphDBMS, and the foreign key of the RDBMS to the relation of the GraphDBMS.

The first table (110) and the second table (120) of the RDBMS are mapped to the first label (410) and the second label (420) of the GraphDBMS. Each record of the RDBMS is mapped to each node of the GraphDBMS.

The first mapping unit (232) forms a relationship between a node of GraphDBMS with an employee ID and a node of GraphDBMS with a department ID based on the relationship between the employee ID and the department ID from the third table composed of reference keys. The relationship defines the relationship and the attributes related to the relationship through commands provided by GraphDBMS. Examples of related attributes include the date of appointment and the job in charge.

The second mapping unit (232) automatically maps the table of the RDBMS inferred by the machine learning unit (220) to the label of the GraphDBMS, the record of the RDBMS to the node of the GraphDBMS, and the foreign key of the RDBMS to the relation of the GraphDBMS.

The second mapping unit (232) can also perform mapping based on a data pattern inferred from data obtained from an application execution result, as in the example of FIG. 3, in the machine learning unit (220).

FIG. 3 illustrates an example of a preferred embodiment of the present invention in which data features are extracted from the execution result data of an application through machine learning in an AI-based data transfer device (200) and then converted into GraphDBMS.

The data conversion unit (240) transfers data from the RDBMS to the GraphDBMS according to the mapping of the first mapping unit (231) and the second mapping unit. The data conversion unit (240) may further include a data verification unit that uses an AI algorithm to identify and correct data inconsistencies or omissions when transferring data from the RDBMS to the GraphDBMS.

The data conversion unit (240) also includes a translation unit that translates SQL queries of the RDBMS into a query language of the GraphDBMS or a Cypher query language.

FIG. 3 illustrates an example of a preferred embodiment of the present invention in which a machine learning unit generates data features between records constituting each of the plurality of tables for data not explicitly defined in the schema of an RDBMS, and infers relationships between the plurality of tables based on the generated data features.

The machine learning unit can infer the relationship between each column (310, 320, 330, 340) of the table in which the data of the application execution result screen is displayed. In the example of Fig. 3, the machine learning unit can infer the relationship between the employee name (310) field and the department name (320) field, and convert Emily (311), Emily (312), ..., Sarah (313), and Chris (314) of the employee name field (310) into nodes (315, 313a, 314a) of GraphDBMS, respectively. In addition, Tech (321), R&D (322), Q&A (323), and Sales (324) of the department name (320) field can also be converted into nodes (321a, 322a, 323a, 324a). In this case, employee-department relationships are created for Emily (311) and Tech (321) and Emily (312) and R&D (322), and since Emily (311, 312) overlap, two employee-department relationships can be created for one Emily (315), Tech (321) and R&D (322).

And, by referring to other fields (330, 340), it can be determined that Emily (315) was assigned to the Tech (321) department on January 3, 2020 and was in charge of Engineer work, and was assigned to the R&D (322) department on March 3, 2022 and was in charge of Developer work.

The methods according to the embodiments of the present invention may be implemented in the form of program commands that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program commands, data files, data structures, etc., either singly or in combination. The program commands recorded on the medium may be those specifically designed and configured for the present invention or may be those known to and usable by those skilled in the art of computer software.

Although the present invention has been described above by means of limited embodiments and drawings, the present invention is not limited to the above embodiments, and those skilled in the art to which the present invention pertains can make various modifications and variations based on this description.

Claims (3)

A preprocessing unit that analyzes the schema of an RDBMS to identify tables, records constituting the tables, and foreign keys indicating relationships between records in each of multiple tables;
A machine learning unit that generates data features between records constituting each of the plurality of tables for data not explicitly defined in the schema of the RDBMS, and infers relationships between the plurality of tables based on the generated data features;
A first mapping unit that maps the table of the RDBMS identified in the above preprocessing unit to the label of the GraphDBMS, the record of the RDBMS to the node of the GraphDBMS, and the foreign key of the RDBMS to the relation of the GraphDBMS, respectively;
A second mapping unit that automatically maps the RDBMS table inferred in the machine learning unit to the GraphDBMS label, the RDBMS record to the GraphDBMS node, and the RDBMS foreign key to the GraphDBMS relation; and
An AI-based data transfer device, characterized by including a data conversion unit that transfers data of an RDBMS to a GraphDBMS according to the mapping of the first mapping unit and the second mapping unit. In paragraph 1,
An AI-based data transfer device characterized by further including a data verification unit that identifies and corrects data inconsistencies or omissions when transferring data from the data conversion unit. In the first paragraph, the data conversion unit
An AI-based data transfer device further comprising a translation unit that translates an SQL query of an RDBMS into a query language of a GraphDBMS or a Cypher query language.

Images