JP7285511B2 - Data management device, data management method, program, and data management system - Google Patents

Description

The present invention relates to a data management device, data management method, program , and data management system.

In recent years, various types of information such as customer information and product information are managed by relational databases (hereinafter simply referred to as "databases"). Even if these databases are originally the same information, they are often individually registered in different databases managed by different administrators. If the databases are different, even if the same information is stored, the configuration of the fields constituting the records in each database, the values stored in each field, and the like may differ.

Therefore, various techniques have been proposed for integrating different databases and mutually referencing them. Among such techniques, there is a technique called "name identification" (see, for example, Patent Document 1).

JP 2012-234343 A

Name identification is a technology that integrates databases based on the similarity of values registered in each database, so it is necessary to access database values when integrating or referencing. For this reason, the entire database must be accessed for name matching, which increases the risk of information leakage. In particular, when confidential information such as personal information is stored in the database, suppression of information leakage becomes a problem.

The present invention has been made in view of these points, and an object of the present invention is to provide a technique for improving security when databases are shared.

A first aspect of the present invention is a data management device. This device includes a vector conversion unit that converts each record constituting a table of a plurality of databases managed by different authority settings into vector data, the vector data, and a data identifier for specifying the vector data. and an administrator identifier for identifying an administrator of a database containing records corresponding to the vector data, and registering them in the vector database; and the administrator identifier registered in the vector database. and a record to be updated in the database managed by the first administrator from the first administrator who is one of the administrators of the plurality of databases an update data acquisition unit that acquires the data identifier of the above, the field name to be updated, and the field information after updating; and the vector data stored in the vector database and the data identifier of the record to be updated are linked. a similarity calculation unit that calculates a similarity to the vector data to be updated, which is the vector data that has been updated; and an update information notification unit for notifying the second administrator of a data identifier for specifying the similar vector data, a field name to be updated, and the updated field information.

The blockchain access unit associates an administrator identifier assigned to each administrator in the vector database with access right information indicating the access right of each administrator in the vector database, and registers the identifier in the blockchain. good too.

The blockchain access unit may refer to the access right information to specify vector data that the first administrator is permitted to access among the vector data registered in the vector database.

The vector conversion unit converts at least updated records among records constituting databases managed by the first manager and the second manager respectively into vector data and registers them in the vector database. good too.

The database access unit may access the vector database to the extent permitted by the access rights acquired from the blockchain by the blockchain access unit.

A second aspect of the present invention is a data management method. In this method, the processor converts each record constituting tables of a plurality of databases each managed with different authority settings into vector data, and the vector data and data for specifying the vector data. a step of associating an identifier with an administrator identifier for specifying an administrator of a database containing records corresponding to the vector data, and registering the administrator identifier registered in the vector database with the administrator identifier; a step of registering in a blockchain; and a data identifier of a record to be updated in a database managed by said first administrator from a first administrator who is one of administrators of said plurality of databases; a step of acquiring a field name to be updated and field information after update; vector data stored in the vector database and vector data in which a data identifier of the record to be updated is linked; a step of calculating a degree of similarity with vector data to be updated; a step of notifying a data identifier for specifying vector data, a field name to be updated, and the updated field information;

A third aspect of the present invention is a program. This program provides a computer with a function of converting each record constituting tables of a plurality of databases each managed with different authority settings into vector data, the vector data, and data for specifying the vector data. A function of associating an identifier with an administrator identifier for identifying an administrator of a database containing records corresponding to the vector data and registering the administrator identifier in the vector database; a function of registering in a blockchain; and a data identifier of a record to be updated in a database managed by said first administrator from a first administrator who is one of the administrators of said plurality of databases; , a function of acquiring a field name to be updated and field information after update, and vector data in which the vector data stored in the vector database and the data identifier of the record to be updated are linked. A second administrator, who is an administrator of a database including a function of calculating a similarity to update target vector data and a record corresponding to similar vector data, which is vector data whose similarity satisfies a predetermined condition, is provided with the similarity It realizes a function of notifying a data identifier for specifying vector data, a field name to be updated, and the field information after the update.

In order to provide this program or update part of the program, a computer-readable recording medium recording this program may be provided, or this program may be transmitted via a communication line.

A fourth aspect of the present invention is a data management system. This system includes the data management device described above and the vector database.

Any combination of the above-described components, and expressions of the present invention converted into methods, devices, systems, computer programs, data structures, recording media, etc. are also effective as aspects of the present invention.

According to the present invention, it is possible to improve security when a database is shared.

1 is a diagram for explaining an overview of a data management device according to an embodiment; FIG. 1 is a diagram schematically showing the functional configuration of a data management device according to an embodiment; FIG. FIG. 4 is a diagram schematically showing an example of the data structure of a database according to the embodiment; FIG. 4 is a diagram schematically showing the data structure of a vector database according to the embodiment; FIG. It is a figure which shows typically the structure of the block chain which concerns on embodiment. 4 is a flowchart for explaining the flow of data management processing executed by the data management device according to the embodiment;

<Overview of Embodiment>
FIG. 1 is a diagram for explaining an outline of a data management device 1 according to an embodiment. In FIG. 1, the first database D1 managed by the first manager M1 and the second database D2 managed by the second manager M2 are configured to communicate with the data management device 1 via the communication network N1. Connected. Since the first database D1 and the second database D2 are managed by different administrators M, the first database D1 and the second database D2 can be said to be databases managed by different authority settings. The data management device 1 is communicably connected to the block chain B via the communication network N1, and is communicatively connected to the vector database V via the secure network N2, which is a dedicated line in which security is ensured. are doing. A data management system S is composed of the data management device 1 and the vector database V. FIG.

Here, Blockchain B functions as a distributed database, and any computer participating in Blockchain B can freely browse the contents. On the other hand, the vector database V is subject to strict management of authority for viewing, and can be accessed only by the data management device 1 via the secure network N2, which is a dedicated line. Hereinafter, when the first manager M1 and the second manager M2 are not distinguished from each other, they are simply referred to as the manager M. Similarly, when the first manager M1 and the second manager M2 are not distinguished from each other, they are simply referred to as database D.

Hereinafter, the procedure of the processing performed by the data management device 1 according to the embodiment will be described by (1) to (8) with reference to FIG. ).

(1) The data management device 1 acquires each record that constitutes the table of the first database D1.
(2) The storage unit 2 acquires each record that constitutes the table of the second database D2.

(3) The data management device 1 converts each acquired record into vector data. This vector data is generated by reflecting the values stored in the fields that make up the record. That is, the more similar the values stored in the fields that form two different records, the more similar each vector data after conversion. On the other hand, it is difficult to restore the information stored in the original field from vector data. The data management device 1 deletes the record after converting it into vector data and does not hold the record itself. A vector data conversion method will be described later.

(4) The data management device 1 associates the vector data obtained by converting the record with a data identifier for identifying the vector data and an administrator identifier for identifying the administrator of the database containing the record. and register it in the vector database V. As a result, the information stored in D1 and the information stored in the second database D2 are stored in the vector database V in the form of vector data. By calculating the degree of similarity between the vector data stored in the vector database V, the data management device 1 can estimate the degree of similarity between the values stored in the fields that form the records that are the basis of the data. can.

(5) The data management device 1 associates the access right information indicating the access right of each administrator with respect to the vector database V with the administrator identifier and registers it in the block chain B via the communication network N1. The access right information is information that defines the permitted range of access to vector data for each administrator M. FIG. For example, manager M can freely access vector data related to database D managed by him/herself. On the other hand, the manager M cannot access the vector data related to the database D managed by the other manager M outside the range approved by the other manager M.

(6) The data management device 1 acquires a data identifier for specifying a record to be updated in the first database D1, a field name to be updated, and field information after update from the first manager M1. .
(7) The data management device 1 converts the record to be updated into vector data, and among the vector data stored in the vector database V, the vector data permitted to be accessed by the first manager M1. Calculate the similarity with

(8) The data management device 1 allows the second manager M2 to access a data identifier for identifying vector data whose degree of similarity satisfies a predetermined condition, a field name to be updated, and field information after update. terminal. Hereinafter, in the present specification, it is assumed that an administrator terminal (not shown) to which the administrator M can access some information is premised on the existence of the administrator terminal, and simply "notify the administrator M" or the like is used. Describe using expressions.

Here, the "predetermined condition" is a condition for satisfying a predetermined degree of similarity in order to consider that the record of the first database D1 and the record of the second database D2 are records storing the same information. be. A specific example of the predetermined condition is exceeding a threshold value for identity determination defined for similarity. In this case, when the degree of similarity between two different vector data exceeds the same determination threshold, the data management device 1 regards the records corresponding to the vector data as records storing the same information.

Thereby, when the record of the first database D1 is changed by the first manager M1, the second manager M2 can correct the contents of the record of the second database D2 which is considered to correspond to the record. That is, the first manager M1 and the second manager M2 can share the contents of the records of the first database D1 and the contents of the records of the second database D2.

Since the data management device 1 stores vector data converted from the records constituting the database D in the vector database V, the data management device 1 stores the vector data in a database accessible via a public communication line such as the Internet. Therefore, the risk of leakage of vector data can be reduced. Also, even if the vector data leaks to a third party, it would be difficult for the third party to restore the original record from the vector data.

Thus, the data management device 1 according to the embodiment can improve security when the database D is shared. Furthermore, information can be linked without integrating the data identifier in the database D managed by each manager M with the data identifier in the database D managed by another manager M different from the manager M. As a result, the data management device 1 according to the embodiment can improve productivity when managing the database D. FIG.

Although FIG. 1 shows a case where there are two databases D, that is, a first database D1 and a second database D2, the number of databases D is not limited to two, and may be three or more.

<Functional Configuration of Data Management Device 1 According to Embodiment>
FIG. 2 is a diagram schematically showing the functional configuration of the data management device 1 according to the embodiment. A data management device 1 includes a storage unit 2 and a control unit 3 . In FIG. 2, arrows indicate main data flows, and data flows not shown in FIG. 2 may exist. In FIG. 2, each functional block does not show the configuration in units of hardware (apparatus), but the configuration in units of functions. Therefore, the functional blocks shown in FIG. 2 may be implemented within a single device, or may be implemented separately within a plurality of devices. Data exchange between functional blocks may be performed via any means such as a data bus, network, or portable storage medium.

The storage unit 2 includes a ROM (Read Only Memory) that stores the BIOS (Basic Input Output System) of the computer that implements the data management device 1, a RAM (Random Access Memory) that serves as a work area for the data management device 1, an OS ( (Operating System), application programs, and a large-capacity storage device such as an HDD (Hard Disk Drive) or an SSD (Solid State Drive) that stores various information referred to when the application program is executed.

The control unit 3 is a processor such as a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit) of the data management device 1, and executes a program stored in the storage unit 2 to perform a vector conversion unit 30, a database access unit 31 , a blockchain access unit 32 , an update data acquisition unit 33 , a similarity calculation unit 34 , and an update information notification unit 35 .

Note that FIG. 2 shows an example in which the data management device 1 is composed of a single device. However, the data management device 1 may be realized by calculation resources such as a plurality of processors and memories, such as a cloud computing system. In this case, each unit that configures the control unit 3 is implemented by executing a program by at least one of a plurality of different processors.

The vector conversion unit 30 acquires each record that constitutes tables of a plurality of databases managed by different administrators. The control unit 3 converts each acquired record into vector data.

FIG. 3 is a diagram schematically showing an example of the data structure of database D according to the embodiment. Specifically, FIG. 3 is a diagram showing the data structure of a member information database for registering member information. A record vector data conversion method executed by the vector conversion unit 30 will be described below with reference to FIG.

The database D is configured by collecting a plurality of records. Here, a record is a unit of information stored in the database D. FIG. When the database D is a member information database, information for one member is stored in each record. As shown in FIG. 3, each record consists of multiple fields. In the example shown in FIG. 3, each record has "data identifier", "last name (kanji)", "first name (kanji)", "last name (hiragana)", "first name (hiragana)", "gender", and "age". , "Address", etc.

Even a member information database for registering member information may have a different structure if the manager M who manages the database is different. For example, in the example shown in FIG. 3, the names of members are registered using "hiragana", but in other databases, "katakana" or "romaji" may be used for registration. Also, in the example shown in FIG. 3, the member's gender is registered using kanji, but in other databases, there are cases where males are represented by symbols such as "M" and females are represented by symbols such as "F". Furthermore, when a member identifier is assigned to each member and managed, it is generally considered that the identifier will be different if the person assigning the identifier is different.

The vector conversion unit 30 according to the embodiment converts each record into vector data in units of records constituting the database D. FIG. Specifically, the vector conversion unit 30 arranges the values of each field included in the obtained record to generate one character string. In the example shown in FIG. 3, when converting a record whose data identifier is DID00001 into vector data, the vector conversion unit 30 removes the data identifier and converts the character string "Taro Yamada Yamada Taro Otoko 32 Tokyo..." to generate

Subsequently, the vector conversion unit 30 converts the generated character string into vector data using a known method known as Word2Vec. Word2Vec has several methods, all of which are realized by neural networks, although detailed description is omitted because it is a known technology. Many character strings (in this embodiment, character strings generated by arranging the values of each field included in a record) are learned using Word2Vec as learning data. A learning model that outputs vector data can be generated. The vector conversion unit 30 converts each record into vector data by using a learning model generated in advance. As an example, although not limited, the vector conversion unit 30 converts each record into 256-bit bit string vector data.

The database access unit 31 receives the vector data converted by the vector conversion unit 30, a data identifier for identifying the vector data, and an administrator identifier for identifying the administrator of the database containing the record corresponding to the vector data. Link it and register it in the vector database. Note that the data identifier for specifying vector data may be the same as the data identifier for specifying the record that is the source of the vector data.

FIG. 4 is a diagram schematically showing the data structure of the vector database V according to the embodiment. As shown in FIG. 4, the vector database V stores, for each administrator identifier, vector data derived from the database D managed by the administrator M identified by the administrator identifier in association with the data identifier. . FIG. 4 shows vector data generated from records in the database D managed by the administrator M whose administrator identifier is MID00001. As shown in FIG. 4, each vector data is assigned a data identifier. For example, vector data whose data identifier is DID00001 is "11001101010010...".

The blockchain access unit 32 registers the administrator identifier registered in the vector database V in the blockchain B. FIG.

FIG. 5 is a diagram schematically showing the configuration of Blockchain B according to the embodiment. As shown in FIG. 5, Blockchain B is information in which a plurality of blocks are linked. Each of the multiple blocks includes a hash value of information indicating the immediately preceding block, an administrator identifier, access right information, and a nonce value. A nonce value is information used to ensure that the hash value of a block meets certain conditions. It should be noted that, as described above, the access right information is information that defines the permitted range of access to vector data for each administrator M. FIG. Therefore, the access right information includes information on data identifiers to which access is permitted.

As shown in FIG. 5, the blockchain access unit 32 registers only the administrator identifier in the blockchain B and does not register the vector data itself. This is because Blockchain B is a public database. As a result, the data management device 1 can prevent vector data from leaking from the block chain B. FIG.

The update data acquisition unit 33 obtains records to be updated in the first database D1 managed by the first manager M1 from the first manager M1 who is one of the managers M of the plurality of databases. , the field name to be updated, and the updated field information.

Based on the data identifier acquired by the update data acquisition unit 33, the similarity calculation unit 34 reads the vector data associated with the data identifier from the vector database V as update target vector data. The similarity calculator 34 calculates the similarity between the vector data to be updated and other vector data stored in the vector database V. FIG. The similarity calculation unit 34 calculates the similarity between vector data using known techniques such as cosine similarity, Euclidean distance, and libraries such as gensim.

The update information notification unit 35 is the second administrator M of the second database D2 that includes records corresponding to similar vector data, which is vector data whose similarity calculated by the similarity calculation unit 34 satisfies a predetermined condition. M2 is notified of the data identifier for specifying similar vector data, the field name to be updated, and the updated field information. Hereinafter, the data identifier for specifying the similar vector data notified to the second manager M2 by the update information notification unit 35, the field name to be updated, and the field information after update will be collectively referred to as "update information". is sometimes described.

The second administrator M2 can specify the record to be updated in the second database D2 based on the data identifier notified from the update information notifier 35. FIG. Further, the second administrator M2 can estimate the field to be updated based on the field name to be updated notified from the update information notification unit 35 . Therefore, the second administrator M2 can rewrite the information of the field to be updated with the information of the field notified by the update information notification unit 35 .

As shown in FIG. 5, the blockchain access unit 32 associates the administrator identifier assigned to each administrator in the vector database V with the access right information indicating the access right of each administrator in the vector database to access the blockchain. Register for B. The manager M is permitted to access the vector data stored in the vector database V only within the scope permitted by the access right information registered in the block chain B. In other words, an administrator M can access vector data derived from databases D managed by other administrators M only when permitted by the other administrator M.

The blockchain access unit 32 refers to the access right information, and the first manager M1 who is the manager of the second database D2 including the record to be updated among the vector data registered in the vector database V Identify the vector data you are authorized to access. The database access unit 31 accesses the vector data stored in the vector database V within the range permitted by the access rights that the blockchain access unit 32 acquires from the block chain B. FIG. In this way, by managing access rights to the vector database V for each manager M using access right information, the data management device 1 can prevent vector data from leaking to the outside.

After the update information notification unit 35 notifies the second manager M2 of the information about the update, the vector conversion unit 30 updates the first database D1 managed by the first manager M1 and the second database managed by the second manager M2. At least the updated records among the records constituting D2 are converted into vector data. The vector conversion unit 30 registers the converted vector data in the vector database V again. Thereby, the data management device 1 can keep the information stored in the vector database V up-to-date.

<Processing Flow of Data Management Method Executed by Data Management Apparatus 1>
FIG. 6 is a flowchart for explaining the flow of data management processing executed by the data management device 1 according to the embodiment. The processing in this flowchart starts, for example, when the data management device 1 is activated.

The vector conversion unit 30 acquires each record that constitutes tables of a plurality of databases D managed by different administrators M (S2). The vector conversion unit 30 converts each acquired record into vector data (S4).

The database access unit 31 stores the vector data converted by the vector conversion unit 30, a data identifier for identifying the vector data, and an administrator identifier for identifying an administrator M of the database D containing records corresponding to the vector data. are linked and registered in the vector database V (S6).

The update data acquisition unit 33 acquires information on field updates from the first administrator M who is one of the administrators M of the plurality of databases D (S8). The similarity calculator 34 calculates the similarity between the vector data stored in the vector database V and the vector data to be updated (S10).

The update information notification unit 35 is the second administrator M of the second database D2 that includes records corresponding to similar vector data, which is vector data whose similarity calculated by the similarity calculation unit 34 satisfies a predetermined condition. M2 is notified of the information regarding the update of the field (S12). When the update information notification unit 35 notifies the second manager M2 of the information regarding the update, the processing in this flowchart ends.

<Effects of the data management device 1 according to the embodiment>
As described above, according to the data management device 1 according to the embodiment, it is possible to improve security when sharing a database.

Although the present invention has been described above using the embodiments, the technical scope of the present invention is not limited to the scope described in the above embodiments, and various modifications and changes are possible within the scope of the gist thereof. be. For example, all or part of the device can be functionally or physically distributed and integrated in arbitrary units. In addition, new embodiments resulting from arbitrary combinations of multiple embodiments are also included in the embodiments of the present invention. The effect of the new embodiment caused by the combination has the effect of the original embodiment.

<Modification>
In the above description, the case where the database D is the member database has been mainly described. However, the database D is not limited to the member database. , database for logistics management and delivery management, database for storing property information and household goods information, database for managing contracts such as insurance and securities, personal behavior history (e.g., biometric information such as electrocardiogram) , activity information such as the number of steps taken), behavior data of IoT (Internet Of Things) devices, and the like. A personal action history can be obtained by, for example, each user always wearing a wearable device equipped with a biosensor.

Here, the biometric information of an individual can be said to be data representing characteristics for distinguishing the individual from others. For example, one person's heartbeat pattern will be different than another person's heartbeat pattern. Therefore, if the database D is a database that stores the behavior history of an individual, when the vector conversion unit 30 converts each record of the database D into vector data, the obtained vector data is data that reflects characteristics that differ for each individual. It is considered that

For example, if biometric information is obtained by having someone wear a wearable device without changing the person's name, gender, etc., the current The degree of similarity between the current vector data and past vector data indicates dissimilarity. In this way, the similarity calculation unit 34 calculates the similarity between the current vector data and the past vector data for a given individual, thereby estimating whether the wearable device is continuously worn by the given individual. can do.

In addition, when the database D stores an individual's action history, the second manager M2 may store a certain individual's action history registered in the first database D1 managed by the first manager M1 as the first manager. 2 It is also conceivable to analyze what kind of action history is registered in the second database D2 managed by the manager M2. In such a case, the second administrator M2 only estimates the record to be updated and the field to be updated in the second database D2 based on the data identifier notified from the update information notification unit 35. The information of the field to be updated is not rewritten to the information of the field notified by the update information notification unit 35 .

Reference Signs List 1 Data management device 2 Storage unit 3 Control unit 30 Vector conversion unit 31 Database access unit 32 Block chain access unit 33 Update data acquisition unit 34 ... similarity calculation unit 35 ... update information notification unit B ... block chain D ... database D1 ... first database D2 ... second database N1 ... communication network N2 ... Secure Network V: Vector database

Claims (7)

A data management device,
a vector conversion unit that is connected to the data management device in a communicable manner and converts each record constituting a table of a plurality of databases managed by different administrators into vector data;
The vector data, the data identifier for identifying the vector data, and the administrator identifier for identifying the administrator of the database containing the record corresponding to the vector data are associated with each other, and the data management device and the dedicated a database access unit for registering in a vector database connected in a communicable manner via a line ;
a blockchain access unit that associates an administrator identifier assigned to each administrator in the vector database with access right information indicating the access right of each administrator in the vector database and registers the administrator identifier in the blockchain;
A record to be updated by the first administrator in a database managed by the first administrator among the plurality of databases from a first administrator who is one of the administrators of the plurality of databases an update data acquisition unit that acquires the data identifier of, the field name to be updated, and the field information after update;
a similarity calculation unit for calculating a similarity between the vector data stored in the vector database and the vector data to be updated, which is vector data associated with the data identifier of the record to be updated;
A data identifier for specifying the similar vector data and a data identifier for specifying the similar vector data and the update target data to the second administrator who is the administrator of the database containing the records corresponding to the similar vector data, which is the vector data whose similarity satisfies a predetermined condition. an update information notification unit that notifies the field name and the updated field information;
A data management device comprising: The blockchain access unit refers to the access right information to identify vector data that the first administrator is permitted to access among the vector data registered in the vector database.
The data management device according to claim 1 . The vector conversion unit converts at least updated records among records constituting databases managed by the first manager and the second manager respectively into vector data and registers the vector data in the vector database.
3. The data management device according to claim 1 or 2. The database access unit accesses the vector database to the extent permitted by the access rights acquired from the blockchain by the blockchain access unit.
The data management device according to any one of claims 1 to 3 . The processor of the data management device
Each record constituting a table of a plurality of databases connected in a manner communicable with the data management device and managed by different administrators , and connected in a manner communicable with the data management device , respectively into vector data; and
The vector data, the data identifier for identifying the vector data, and the administrator identifier for identifying the administrator of the database containing the record corresponding to the vector data are associated with each other, and the data management device and the dedicated registering in a vector database communicatively connected via a line ;
a step of linking an administrator identifier assigned to each of the administrators in the vector database with access right information indicating each administrator's access rights in the vector database and registering the administrator identifier in the blockchain;
A record to be updated by the first administrator in a database managed by the first administrator among the plurality of databases from a first administrator who is one of the administrators of the plurality of databases a step of obtaining a data identifier of, a field name to be updated, and updated field information;
calculating the degree of similarity between the vector data stored in the vector database and the vector data to be updated, which is vector data associated with the data identifier of the record to be updated;
A data identifier for specifying the similar vector data and a data identifier for specifying the similar vector data and the update target data to the second administrator who is the administrator of the database containing the records corresponding to the similar vector data, which is the vector data whose similarity satisfies a predetermined condition. a step of notifying the field name and the updated field information;
data management methods that perform to the computer,
A function of converting each record constituting a table of a plurality of databases connected to the computer in a communicable manner and managed by different administrators into vector data;
The vector data, a data identifier for identifying the vector data, and an administrator identifier for identifying an administrator of a database containing records corresponding to the vector data are linked to connect the computer and the dedicated line. a function to register in a vector database connected in a communicable manner via
A function of linking an administrator identifier assigned to each of the administrators in the vector database with access right information indicating the access rights of each administrator in the vector database and registering the identifier in the blockchain;
A record to be updated by the first administrator in a database managed by the first administrator among the plurality of databases from a first administrator who is one of the administrators of the plurality of databases A function to acquire the data identifier of, the field name to be updated, and the field information after updating,
a function of calculating the degree of similarity between the vector data stored in the vector database and the vector data to be updated, which is vector data associated with the data identifier of the record to be updated;
A data identifier for specifying the similar vector data and a data identifier for specifying the similar vector data and the update target data to the second administrator who is the administrator of the database containing the records corresponding to the similar vector data, which is the vector data whose similarity satisfies a predetermined condition. a function of notifying the field name and the updated field information;
program to realize a data management device according to any one of claims 1 to 4 ;
the vector database;
A data management system with

Images