Transfer destination determination device, transfer system, transfer destination determination method, transfer method, and program

Abstract

In order to determine a folder to which a file is transferred between file servers having different folder attribute systems, this transfer destination determination device includes: a teacher data creation means that creates teacher data using a learning file; a learning means that generates, by teacher-equipped machine learning that uses teacher data, a transfer destination determination model that outputs, in response to an input of file properties of a transfer origin folder of a transfer origin file server, information related to a transfer destination folder of a transfer destination file server that is the transfer destination; and a transfer destination determination means that uses the transfer destination determination model to determine the transfer destination folder of the transfer destination file server, that is the transfer destination of the file of the transfer origin folder of the transfer origin file server.

Description

Technical Field

[0001] The present invention relates to a technique for transferring files.

Background Art

[0002] Patent Document 1 describes a file sharing system that stores files with different attributes.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] When the folder attribute systems are different among a plurality of file servers, it is necessary to determine the folder to which a file is to be transferred during file transfer. The technique described in Patent Document 1 is a technique for setting access information of a file after migration, and is not a technique for determining the folder to which a file is to be transferred.

[0005] An object of the present invention is to provide a transfer destination determination device or the like that determines a folder to which a file is to be transferred among file servers having different folder attribute systems.

Means for Solving the Problems

[0006] A learning device according to an aspect of the present invention includes: teacher data creation means for creating teacher data using learning files; learning means for generating a transfer destination determination model that outputs information regarding a transfer destination folder of a transfer destination file server according to an input of properties of files in a transfer source folder of a transfer source file server by supervised machine learning using the teacher data; and includes.

[0007] In one embodiment of the present invention, the destination determination device uses a destination determination model that outputs information about the destination folder of the destination file server, which is the destination of the file in the source folder of the source file server, to determine the destination folder of the destination file server, which is the destination of the file in the source folder of the source file server. Destination determination means including.

[0008] In one embodiment of the present invention, the destination determination device Teacher data creation means for creating teacher data using learning files, Learning means for generating a destination determination model that outputs information about the destination folder of the destination file server, which is the destination, in response to the input of the properties of the file in the source folder of the source file server, by supervised machine learning using the teacher data, Destination determination means for determining the destination folder of the destination file server, which is the destination of the file in the source folder of the source file server, using the destination determination model, and including.

[0009] In one embodiment of the present invention, the transfer system The above destination determination device, A file information acquisition device for generating learning files, A file transfer device that transfers the file in the source folder of the source file server to the destination folder of the destination file server determined by the destination determination device, and including.

[0010] In one embodiment of the present invention, the destination determination method Create teacher data using learning files, Generate a destination determination model that outputs information about the destination folder of the destination file server, which is the destination, in response to the input of the properties of the file in the source folder of the source file server, by supervised machine learning using the teacher data, Using the transfer destination determination model, determine the transfer destination folder of the transfer destination file server for the files in the transfer source folder of the transfer source file server.

[0011] The transfer method in one embodiment of the present invention is The transfer destination determination device executes the above transfer destination determination method, The file information acquisition device generates a learning file, The file transfer device transfers the files in the transfer source folder of the transfer source file server to the transfer destination folder of the transfer destination file server determined by the transfer destination determination device.

[0012] The recording medium in one embodiment of the present invention is A process of creating teacher data using a learning file, A process of generating a transfer destination determination model that outputs information regarding the transfer destination folder of the transfer destination file server according to the input of the properties of the files in the transfer source folder of the transfer source file server by supervised machine learning using the teacher data, A process of determining the transfer destination folder of the transfer destination file server for the files in the transfer source folder of the transfer source file server using the transfer destination determination model Record a program that causes a computer to execute.

Advantages of the Invention

[0013] According to the present invention, it is possible to reduce the load of determining the folder to which a file is transferred between file servers having different folder attribute systems.

Brief Description of the Drawings

[0014]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Mode for Carrying Out the Invention

[0015] <Related Art> Cloud-based online file servers can often reduce additional investment and operation burdens compared to on-premises file server environments. Therefore, the demand for transferring files from on-premises file server environments to cloud-based online file server environments is increasing. Hereinafter, the cloud-based online file server environment is referred to as "cloud". Also, the on-premises file server environment is referred to as "on-premises".

[0016] However, generally, the system of on-premises file and folder attributes often differs from the system of cloud file and folder attributes. Thus, when the system of folder attributes of the source file server is different from the system of folder attributes of the destination file server, files cannot be transferred while maintaining the folder configuration. If the folder configuration cannot be maintained, it is necessary to determine the folder from which to transfer the files among the folders of the destination file server.

[0017] As a method for determining the folder from which to transfer files, for example, non-hierarchical cluster analysis is used. However, when using non-hierarchical cluster analysis, it is necessary to appropriately set parameters. And generally, users and the like analyze the actual folders and adjust the parameters. Thus, the determination of the folder from which to transfer the current files requires parameter adjustment work by users and the like, and requires a lot of man-hours. Also, ensuring objectivity in determining parameters based on human judgment is difficult.

[0018] As described below, each embodiment of the present invention automates the parameter adjustment work, determines appropriate parameters, and reduces the man-hours of users and the like. Hereinafter, the embodiments of the present invention will be described with reference to the drawings. However, each embodiment of the present invention is not limited to the description in each drawing. Also, the embodiments can be combined as appropriate. In the following description, to clarify the distinction between folders, the folders of the source file server may be referred to as "source folders", and the folders of the destination file server may be referred to as "destination folders".

[0019] <First Embodiment> FIG. 1 is a diagram showing an example of the configuration of the transfer destination determination device 4 according to the first embodiment of the present invention. The transfer destination determination device 4 includes a teacher data creation unit 5, a learning unit 6, and a transfer destination determination unit 7. The teacher data creation unit 5 creates teacher data using a learning file. The learning unit 6 generates a transfer destination determination model that outputs information regarding a transfer destination folder of a transfer destination file server in response to an input of properties of files in a source folder of the transfer destination file server by supervised machine learning using the teacher data. The transfer destination determination unit 7 determines a transfer destination folder of a file in a source folder of a source file server as a transfer destination using the transfer destination determination model. Note that a folder is a storage location of a file in a file system, and may be called a directory in some systems.

[0020] That is, the transfer destination determination device 4 executes machine learning using teacher data generated using a learning file to generate a transfer destination determination model. Then, the transfer destination determination device 4 determines a transfer destination folder of a file in a source folder of a source file server as a transfer destination using the generated transfer destination determination model. In this way, the transfer destination determination device 4 uses the transfer destination determination model generated by supervised machine learning to determine a folder that is a transfer destination of a file. Therefore, the transfer destination determination device 4 can determine a folder to which a file is to be transferred between file servers having different folder attribute systems without requiring parameter adjustment work by a user or the like. In other words, the transfer destination determination device 4 automates the parameter adjustment work, determines appropriate parameters, and can determine a folder to which a file is to be transferred between file servers having different folder attribute systems. As a result, the transfer destination determination device 4 reduces the user's load in determining a folder. Further, since the transfer destination determination device 4 uses the transfer destination determination model generated by supervised machine learning, it can determine a transfer destination more objectively with respect to manual adjustment.

[0021] Note that the files to be transferred may be only some of the files in the source file server, rather than all of them. In this case, the destination determination device 4 only needs to determine the destination folder for the "files to be transferred", and there is no need to operate on the "files not to be transferred". Therefore, in the following description, for the sake of convenience, the "files to be transferred" in the source folder of the source file server are simply referred to as "files". That is, in the following description, the files in the source file server are files to be transferred unless otherwise specified.

[0022] The destination determination model may select multiple folders instead of one folder for each file as the destination folder of the destination file server. When the destination determination model selects multiple destination folders, the destination determination unit 7 determines the destination folder to be used as the destination based on a predetermined rule from among the multiple destination folders. Next, an example of the rule used by the destination determination unit 7 will be described.

[0023] When the destination determination model selects multiple destination folders, the destination determination unit 7 operates as follows. The destination determination unit 7 obtains the confidence level for each file in the source folder of the source file server with respect to each of the destination folders of the destination file server determined using the destination determination model. For example, the destination determination unit 7 obtains the confidence level of the determined folder from the destination determination model. Then, based on the confidence level, the destination determination unit 7 determines the destination folder of the destination file server to which the file in the source folder of the source file server is to be transferred.

[0024] For example, the transfer destination determination unit 7 operates as follows to determine the transfer destination folder based on the confidence level. The transfer destination determination unit 7 calculates the harmonic mean of the confidence levels of the files for each transfer destination folder of the transfer destination file server. Then, the transfer destination determination unit 7 determines the transfer destination folder of the transfer destination file server to which the files in the transfer source folder of the transfer source file server are to be transferred based on the harmonic mean of the confidence levels. For example, the transfer destination determination unit 7 determines the transfer destination folder with the largest harmonic mean as the transfer destination folder.

[0025] The confidence level is a probability indicating how correct the folder determined by the transfer destination determination model is for each file. Also, the reason the transfer destination determination unit 7 uses the harmonic mean is as follows. The harmonic mean is generally an appropriate mean when considering the mean of "rates" or "ratios". And the confidence level is a probability, which is an example of a "rate". Therefore, the transfer destination determination unit 7 uses the harmonic mean as the mean.

[0026] In this way, the transfer destination determination device 4 may determine the folder to which the file is to be transferred using the harmonic mean of the confidence levels of the files for each folder. However, the transfer destination determination unit 7 may use a mean different from the harmonic mean. For example, the transfer destination determination unit 7 may use the arithmetic mean or the geometric mean. Alternatively, the transfer destination determination unit 7 may use a value different from the confidence level in determining the folder.

[0027] Note that the transfer destination determination device 4 may provide the generated transfer destination determination model to other devices. Alternatively, the transfer destination determination device 4 may determine the transfer destination folder using the transfer destination determination model generated by other devices. Next, the transfer destination determination device 4 in these cases will be described.

[0028] FIG. 14 is a block diagram showing an example of the configuration of a learning device 41 that generates a transfer destination determination model. The learning device 41 includes a teacher data creation unit 5 and a learning unit 6. The teacher data creation unit 5 creates teacher data using a file for learning the transfer destination determination unit. The learning unit 6 generates a transfer destination determination model that outputs information regarding a transfer destination folder of a transfer destination file server that is the transfer destination in response to an input of properties of files in a transfer source folder of a transfer source file server by supervised machine learning using the teacher data. Then, the learning unit 6 outputs the transfer destination determination model.

[0029] FIG. 15 is a block diagram showing an example of the configuration of a transfer destination determination device 42 that uses the transfer destination determination model. The transfer destination determination device 42 acquires a transfer destination determination model that outputs information regarding a transfer destination folder of a transfer destination file server that is the transfer destination of a file in a transfer source folder of a transfer source file server. Then, the transfer destination determination device 42 includes a transfer destination determination unit 7 that determines a transfer destination folder of a transfer destination file server that is the transfer destination of a file in a transfer source folder of a transfer source file server using the transfer destination determination model.

[0030] <Second Embodiment> Next, the second embodiment will be described with reference to the drawings. FIG. 2 is a block diagram showing an example of the configuration of a transfer system 10 according to the second embodiment. The transfer system 10 includes a transfer source file server 1, a file information acquisition device 2, a storage device 3, a transfer destination determination device 4, a file transfer device 8, and a transfer destination file server 9.

[0031] The source file server 1 is a file server that serves as the source of file transfer. The destination file server 9 is a file server that serves as the destination of file transfer. And at least a part of the folder attribute system in the source file server 1 is different from the folder attribute system in the destination file server 9. For example, the source file server 1 is an on-premises file server. And the destination file server 9 is a cloud file server. However, each of the source file server 1 and the destination file server 9 is not limited to the above configuration. For example, both the source file server 1 and the destination file server 9 may be on-premises file servers, but file servers with different operating systems from each other. Alternatively, the source file server 1 and the destination file server 9 may be cloud file servers with different cloud systems from each other.

[0032] Note that the folder attribute is the access right to the folder. However, the folder attribute is not limited to the access right. The folder attribute may be at least one of a storage attribute (archive attribute), a compression attribute, and an encryption attribute, or a combination of these and the access right. However, the folder attribute is not limited to the above, and may be an attribute different from the above.

[0033] The file information acquisition device 2 generates a learning file used by the destination determination device 4. The file information acquisition device 2 stores the generated learning file in the storage device 3. The learning file, as data used for machine learning, includes at least the properties of the file correctly transferred and the folder of the transfer destination of the file. For example, the file information acquisition device 2 generates a learning file including the properties of the file correctly transferred and the folder of the transfer destination of the file. For example, the file information acquisition device 2 generates a learning file as follows. Note that the file information acquisition device 2 may generate a learning file using a method different from the following description.

[0034] For example, the file information acquisition device 2 acquires the attributes and configuration of the source folder of the source file server 1, the properties of the files in the source folder, and the attributes and configuration of the destination folder of the destination file server 9. Then, the file information acquisition device 2 extracts files from the information of the files that were correctly transferred in the past and stored in a storage device (not shown), where the file properties are the same as those of the files in the source folder of the source file server 1. When there are multiple file properties, the file information acquisition device 2 extracts the files corresponding to each property.

[0035] Then, the file information acquisition device 2 extracts files from the extracted files where the source folder has the same attributes and configuration as the folder of the source file server 1. Further, the file information acquisition device 2 extracts files from the files extracted as above where the destination folder has the same attributes and configuration as the folder of the destination file server 9. Then, the file information acquisition device 2 combines the properties of the files extracted as above and the destination folder to generate a learning file.

[0036] Note that the file information acquisition device 2 may acquire the properties of the files transferred from the source file server 1 to the destination file server 9 in the past and the transferred destination folders, and create a learning file. Alternatively, the file information acquisition device 2 may acquire necessary information from users or the like and generate a learning file based on the acquired information. Alternatively, the file information acquisition device 2 may acquire a learning file from users or the like.

[0037] A property is information indicating the characteristics of a folder and a file. For example, the properties of a folder may include at least one of the folder name, the type of the folder, the folder in which the folder is stored, the size of the folder, the owner of the folder, the contents of the folder, the creation date and time, the update date and time, and the access date and time. Note that the contents of the folder are the number of files and folders contained in the folder. The contents of the folder may include the total size of the files contained in the folder and the size of the folders contained in the folder. The properties of a file may include at least one of the file name, the type of the file, the program that created the file, the folder in which the file is stored, the size, the owner, the creation date and time, the update date and time, and the access date and time. The properties are not limited to the above and may include information different from the above. For example, the properties may include attributes. For example, the properties of a file may include the access rights of the file.

[0038] The storage device 3 stores information used by the devices included in the transfer system 10. For example, the storage device 3 stores information related to the operations of at least one of the file information acquisition device 2, the transfer destination determination device 4, and the file transfer device 8. For example, the storage device 3 stores information related to the transfer of files when the file transfer device 8 transfers files in the source folder of the source file server 1 to the destination folder of the destination file server 9.

[0039] The file transfer device 8 transfers the files in the source folder of the source file server 1 to the destination folder of the destination file server 9 based on the information related to the transfer of the files determined by the transfer destination determination device 4 and stored in the storage device 3. Therefore, the information related to the transfer of files includes at least information indicating the destination folder that is the transfer destination of the files. Also, the information related to the transfer of files may include other information. For example, the information related to the transfer of files may include other information such as information indicating the files to be transferred.

[0040] The transfer destination determination device 4 determines the transfer destination folder of the transfer destination file server 9 that transfers the files in the transfer source folder of the transfer source file server 1, and stores it in the storage device 3 as information related to the file transfer. Similar to the first embodiment, the transfer destination determination device 4 includes a teacher data creation unit 5, a learning unit 6, and a transfer destination determination unit 7.

[0041] The teacher data creation unit 5 creates teacher data using the properties of the files included in the learning files stored in the storage device 3. The teacher data creation unit 5 stores the created teacher data in the storage device 3. Note that the teacher data creation unit 5 may use all the properties included in the learning files or some of the properties. For example, the teacher data creation unit 5 does not necessarily need to extract the characteristics of the properties that are not used in the machine learning in the learning unit 6 described later.

[0042] The teacher data includes a correct label which is information indicating the correct answer. In the present disclosure, the teacher data includes, as the correct label, information related to the properties of the correct folder determined based at least on the properties of the files. Further, the teacher data includes information related to the properties of the files used for determining the folder. For example, when determining the folder based on the properties related to the owner and the access right, the teacher data includes the information related to the owner and the access right and the information related to the folder. The information included in the teacher data is not limited to the information related to the owner and access right of the file and the information related to the folder, and may be other information.

[0043] The learning unit 6 generates a transfer destination determination model as a result of machine learning by performing supervised machine learning using the teacher data stored in the storage device 3 as input. More specifically, the learning unit 6 generates a transfer destination determination model that outputs information regarding a folder from the properties of a file based on the properties and folders of the files included in the teacher data. Then, the learning unit 6 stores the generated transfer destination determination model in the storage device 3. The transfer destination determination unit 7 determines the transfer destination folder of the transfer destination file server, which is the transfer destination of the files in the transfer source folder of the transfer source file server, using the transfer destination determination model stored in the storage device 3. More specifically, the transfer destination determination unit 7 inputs the properties of the files in the transfer source folder into the transfer destination determination model and determines the folder output by the transfer destination determination model as the transfer destination folder of the file. Then, the transfer destination determination unit 7 stores the determined transfer destination folder in the storage device 3 as information related to the transfer of the file.

[0044] As already described, there may be a case where the transfer destination determination model selects a plurality of transfer destination folders for one file. Even in such a case, the transfer destination determination unit 7 determines one transfer destination folder from the plurality of transfer destination folders determined by the transfer destination determination model for each file. However, when there are a plurality of transfer source folders, for each transfer source folder, the transfer destination folder to which the file is transferred may be a different folder. That is, there may be a case where there are a plurality of transfer destination folders to which the file is transferred. In this case, information on the files to be transferred to each transfer destination folder is required. Therefore, in this case, the transfer destination determination unit 7 associates the information on the plurality of transfer destination folders and the information on the files to be transferred to each transfer destination folder as information related to the transfer of the file and stores it in the storage device 3.

[0045] Next, with reference to the drawings, the operations of each part of the destination determination device 4 will be described. In the following description, as an example of the attributes of a folder, the "access right" of a folder in the file system is used. That is, the access right system of the folders in the source file server 1 is different from the access right system of the folders in the destination file server 9. Specifically, in the source file server 1, a folder with an access right different from that of the top-level folder becomes a folder whose folder configuration cannot be maintained. However, this does not limit the attributes in the second embodiment to access rights.

[0046] Maintaining the folder configuration means that the folder configuration in the destination file server 9 includes at least the same folder configuration as the folder configuration for storing the files to be transferred in the source file server 1. In this case, in addition to the same folder configuration as that of the source file server 1, the destination file server 9 may include folders not included in the folder configuration of the source file server 1.

[0047] First, with reference to the drawings, the operation of the teacher data creation unit 5 will be described. FIG. 3 is a flowchart showing an example of the operation of the teacher data creation unit 5. Before the operation described below, it is assumed that the file information acquisition device 2 generates a learning file based on the information of the files correctly transferred in the past and stores it in the storage device 3. That is, the storage device 3 stores a learning file for creating teacher data.

[0048] The teacher data creation unit 5 acquires the properties of the files included in the learning file stored in the storage device 3 (step S101). Note that the teacher data creation unit 5 may acquire the learning file from the file information acquisition device 2 and acquire the properties from the acquired learning file. Then, the teacher data creation unit 5 extracts the features of the file based on the properties of the file (step S102). Then, the teacher data creation unit 5 creates teacher data using the extracted features of the file (step S103). Then, the teacher data creation unit 5 stores the created teacher data in the storage device 3. Note that the teacher data creation unit 5 may output the created teacher data to the learning unit 6.

[0049] The operation of the teacher data creation unit 5 will be described using specific data. FIG. 4 is a diagram showing an example of the properties of the files included in the learning file. Note that FIG. 4 shows the folders and files in the transfer destination file server 9 of the correctly transferred file. That is, the transfer destination folder of the "file having the same properties as the file shown in FIG. 4" is the folder shown in FIG. 4. In the present embodiment, the owner and access rights of the file are used as the properties of the file used to determine the transfer destination folder. Therefore, the above "file having the same properties as the file shown in FIG. 4" is a file having the same owner and access rights.

[0050] The properties shown in FIG. 4 include the folder of the file, the file name, the owner of the file, and the access rights of each part. For example, the third file from the top in FIG. 4 is a file with the file name "c.xlsx" stored in the folder "￥￥fileServer￥Sales Department￥Query Documents". Hereinafter, the file will be referred to using "file name" or "folder + file name". For example, the third file will be referred to as file "c.xlsx" or file "￥￥fileServer￥Sales Department￥Query Documents￥c.xlsx". And the owner of the file "c.xlsx" is the Sales Department. Furthermore, as the access rights of the file "c.xlsx", "Full Control" is set for the Sales Department and "Read" is set for the Development Department. No access rights are set for the Personnel Department.

[0051] The teacher data creation unit 5 acquires the properties of the files included in the learning files shown in FIG. 4 from the storage device 3. Then, the teacher data creation unit 5 extracts the features of the files from the file properties. The features extracted by the teacher data creation unit 5 may be determined corresponding to the specifications of the file system including the folder and the specifications of the transfer destination determination model used by the user or the like for determining the transfer destination folder in advance and set in the transfer destination determination device 4. For example, the teacher data creation unit 5 may extract the following features. However, the features extracted by the teacher data creation unit 5 are not limited to the features described below.

[0052] The teacher data creation unit 5, as a feature, converts the folder, owner, and access rights among the properties into predetermined scales or numerical values. The scales are not limited to these, but for example, they are nominal scales, ordinal scales, interval scales, or ratio scales. A nominal scale is a scale used simply for distinction. An ordinal scale is a scale with meaning in terms of magnitude relationship. An interval scale is a scale with meaning in terms of the difference between numerical values. A ratio scale is a scale with meaning in terms of both the difference and the ratio of numerical values. In the following description, the teacher data creation unit 5 uses the scales shown in the following (1) to (3). In this embodiment, the "owner", "access rights", and "folder" of the file are used. However, this does not limit this embodiment. This embodiment may execute machine learning using part or all of the "file name", for example.

[0053] (1) Owner The teacher data creation unit 5 divides the "owner" in the property into three nominal scales corresponding to the "Sales Department", "Development Department", and "Personnel Department" respectively. The three nominal scales are nominal scales where each department is "1" when it is the owner and "0" when it is not the owner. For example, as a feature of the third file "c.xlsx" in FIG. 4, the teacher data creation unit 5 sets "1" for the "Sales Department" and "0" for the "Development Department" and the "Personnel Department".

[0054] (2) Access rights The teacher data creation unit 5 converts the "access rights" of each department in the property into a numerical value or an ordinal scale. For example, the teacher data creation unit 5 uses a conversion table to convert the "access rights" into a numerical value or an ordinal scale. FIG. 5 is a diagram showing an example of the conversion table. For example, based on the conversion table in FIG. 5, the teacher data creation unit 5 converts the access rights of the Sales Department of the third file "c.xlsx" in FIG. 4 to "1.0", the access rights of the Development Department to "0.4", and the access rights of the Personnel Department to "0.0".

[0055] (3) "Folder" The teacher data creation unit 5 assigns an integer of 0 or more as a nominal scale to each folder. Specifically, the teacher data creation unit 5 checks the folders of the files in a predetermined order, assigns "0" to the first folder that appears, and then assigns consecutive integers starting from 1 to the newly appearing folders. For example, the teacher data creation unit 5 assigns "0" to the first folder "￥￥fileServer￥Sales Department" from the top in FIG. 4, and assigns "1" to the next folder "￥￥fileServer￥Sales Department￥Inquiry Documents". Furthermore, the teacher data creation unit 5 assigns "2" to the next folder "￥￥fileServer￥Development Department". Note that the transfer destination determination device 4 determines the transfer destination folder. Therefore, the folder becomes the correct label in the teacher data. That is, in the machine learning described later, the machine learning of the transfer destination determination model is executed so that the folder determined using the owner and the access right becomes the folder shown in FIG. 4.

[0056] The teacher data creation unit 5 extracts the features of the files included in the learning files as described above based on the properties. FIG. 6 is a diagram showing an example of the features of the files extracted by the teacher data creation unit 5 based on the properties of the files in FIG. 4. For example, the teacher data creation unit 5 extracts the folder "1" in the third row of FIG. 6, the owner "1, 0, 0", and the access right "1.0, 0.4, 0.0" as the features of the third file "c.xlsx" in FIG. 4. Then, the teacher data creation unit 5 stores the extracted file features in the storage device 3 as teacher data. For example, the teacher data creation unit 5 stores the features in FIG. 6 as teacher data.

[0057] Next, with reference to the drawings, the operation of the learning unit 6 will be described. FIG. 7 is a flowchart showing an example of the operation of the learning unit 6. The learning unit 6 acquires teacher data, that is, the features extracted by the teacher data creation unit 5, from the storage device 3 (step S201). Note that the learning unit 6 may acquire teacher data from the teacher data creation unit 5. Then, the learning unit 6 executes supervised machine learning using the acquired teacher data to generate a transfer destination determination model that outputs information related to the folder to which the file is to be transferred (step S202). The learning unit 6 generates a transfer destination determination model by general machine learning. General machine learning includes a support vector machine, naive Bayes, or a neural network, etc. However, the learning unit 6 may generate a transfer destination determination model by other machine learning. Thus, since machine learning is used, the learning unit 6 can generate a transfer destination determination model with parameters adjusted without the need for parameter adjustment work by users or the like. Then, the learning unit 6 stores the generated transfer destination determination model in the storage device 3 (step S203). Note that the learning unit 6 may output the generated transfer destination determination model to the transfer destination determination unit 7.

[0058] Next, with reference to the drawings, the operation of the transfer destination determination unit 7 will be described. However, the following description is an explanation of the operation of the transfer destination determination unit 7 when the transfer destination determination model selects a plurality of transfer destination folders. When the transfer destination determination model selects one transfer destination folder, the transfer destination determination unit 7 may simply store the transfer destination folder selected by the transfer destination determination model as information related to file transfer.

[0059] FIG. 8 is a diagram showing the folder configuration in the source file server 1 used for explaining the operation. In FIG. 8, "Folder A" is the top-level folder. Hereinafter, the access right of "Folder A" is referred to as access right A. Note that the top-level folder may also be called the "root folder". Further, "Folder A" is a folder that stores "Folder B" and "Folder C". A folder that stores folders may also be called the "parent folder" of the stored folders. That is, "Folder A" is the parent folder of "Folder B" and "Folder C". Conversely, "Folder B" and "Folder C" are lower-level folders directly stored in "Folder A". Lower-level folders directly stored may also be called "child folders". That is, "Folder B" and "Folder C" are the child folders of "Folder A". Similarly, "Folder C" is the parent folder of "Folder D" and "Folder E". Also, "Folder D" and "Folder E" are the child folders of "Folder C".

[0060] And, "Folder B" is an inherited folder. That is, the access right of "Folder B" is the same access right A as the access right A of the parent folder "Folder A". "Folder C" is a non-inherited folder. That is, the access right of "Folder C" is a different access right from the access right A of the parent folder "Folder A". Hereinafter, the access right of "Folder C" is referred to as access right B. "Folder D" is an inherited folder. That is, the access right of "Folder D" is the same access right B as the access right B of the parent folder "Folder C". "Folder E" is a non-inherited folder. That is, the access right of "Folder E" is a different access right from the access right B of the parent folder "Folder C". Hereinafter, the access right of "Folder E" is referred to as access right C.

[0061] In FIG. 8, the inherited folder and the non-inherited folder are each one, but this is just an example. At least one of the inherited folder and the non-inherited folder may be plural. Also, in FIG. 8, the number of sub-folders is an example. Although not shown, at least one of Folder A to Folder E may further include sub-folders. In this case, the sub-folders not shown may further include sub-folders. Thus, the number of folder hierarchies can be set as appropriate and is not limited. Therefore, in the following description, the folders subordinate to a folder may sometimes be collectively referred to as "subordinate folders".

[0062] The transfer destination determination unit 7 determines, as the transfer destination folder for transferring the files of the folder whose folder configuration cannot be maintained among the folders of the source file server 1 shown in FIG. 8, any folder in the destination file server 9. Here, "Folder C" is a non-inherited folder. Therefore, "Folder C" and "the folders subordinate to Folder C" are non-inherited folders with respect to "Folder A", and are folders that cannot transfer files to the destination file server 9 while maintaining the folder configuration in the source file server 1. Therefore, the file transfer device 8 cannot transfer the files of "Folder C" and "the folders subordinate to Folder C" to the destination file server 9 while maintaining the folder configuration in the source file server 1. In the destination file server 9 and the source file server 1, the folders maintaining the same configuration are the root folder "Folder A" and its inherited folder "Folder B". Therefore, the file transfer device 8 transfers the files stored in "Folder C" and "the folders subordinate to Folder C" of the source file server 1 to either "Folder A" or "Folder B" in the destination file server 9.

[0063] That is, the destination determination unit 7 of the destination determination device 4 determines either "Folder A" or "Folder B" of the destination file server 9 as the transfer destination of the files to be transferred in "Folder C" of the source file server 1 and "folders under Folder C". Hereinafter, as an example, the destination determination unit 7 determines either "Folder A" or "Folder B" of the destination file server 9 as the transfer destination of file X included in "Folder C" which is non-inherited in the source file server 1. Further, the destination determination unit 7 determines either "Folder A" or "Folder B" of the destination file server 9 as the transfer destination of file Y included in "Folder D" which is "a folder under Folder C".

[0064] Figure 9 is a flowchart showing an example of the operation of the destination determination unit 7. The destination determination unit 7 acquires a destination determination model from the storage device 3 (step S301). Note that the destination determination unit 7 may acquire the destination determination model from the learning unit 6. Then, the destination determination unit 7 repeats loop L1 for each non-inherited folder including the file to be transferred (step S302).

[0065] Repeating loop L1 for each non-inherited folder specifically means repeating loop L1 for each "non-inherited folder and its sub-folders" where the folders do not overlap with each other. When there is one non-inherited folder containing the files to be transferred, the transfer destination determination unit 7 executes loop L1 for that non-inherited folder. When there are multiple non-inherited folders containing the files to be transferred, the transfer destination determination unit 7 extracts a non-inherited folder that does not contain a non-inherited folder as the upper-level folder. That is, the transfer destination determination unit 7 extracts the topmost folder as non-inherited. Then, the transfer destination determination unit 7 repeats loop L1 with the extracted "non-inherited folder and its sub-folders" as one unit. In this way, the transfer destination determination unit 7 processes the "non-inherited folder and its sub-folders" together. Therefore, in the following description, the "non-inherited folder and its sub-folders" may also be collectively referred to as the "non-inherited folder". In this way, the transfer destination determination unit 7 repeats loop L1 for each non-inherited folder that contains the files to be transferred and where the folders do not overlap with each other.

[0066] In loop L1, the transfer destination determination unit 7 repeats loop L2 for each file to be transferred included in the non-inherited folder (step S303). In loop L2, the transfer destination determination unit 7 determines the transfer destination folder for each file by applying the transfer destination determination model. However, as described above, the transfer destination determination model selects multiple folders. Specifically, the transfer destination determination model selects folder A and folder B for each of file X and file Y. Therefore, the transfer destination determination unit 7 obtains the confidence level of each folder determined for each file from the transfer destination determination model.

[0067] Specifically, the transfer destination determination unit 7 obtains, from the transfer destination determination model, the confidence levels for the determined folder A and folder B for each of file X and file Y. That is, the transfer destination determination unit 7 obtains, for all the files to be transferred included in the folder, the transfer destination folder and the file confidence level for the folder (step S304). Then, the transfer destination determination unit 7 stores the obtained transfer destination folder and the file confidence level for the folder in the storage device 3.

[0068] When loop L2 ends, that is, when the transfer destinations and confidence levels of all the files to be transferred included in the non-inherited folder are stored in the storage device 3, the transfer destination determination unit 7 determines the transfer destination folder of the file based on the file confidence level (step S305). Specifically, the transfer destination determination unit 7 calculates the harmonic mean of the file confidence levels for each transfer destination folder. Then, the transfer destination determination unit 7 determines the folder with the maximum harmonic mean as the transfer destination folder of the file. Then, the transfer destination determination unit 7 stores the determined folder in the storage device 3 as information related to the transfer of the file.

[0069] FIG. 10 is a diagram showing an example of the confidence level and its harmonic mean. In FIG. 10, the confidence level of folder A is 10% for file X and 40% for file Y. As a result, the harmonic mean of the confidence level of folder A is 16%. Similarly, the harmonic mean of the confidence level of folder B is 24%. As a result, the transfer destination determination unit 7 determines folder B as the transfer destination of file X and file Y. Then, the transfer destination determination unit 7 stores folder B in the storage device 3 as information related to the transfer of file X and file Y. The transfer destination determination unit 7 repeats the above operations for each non-inherited folder that is the processing unit of loop L1. Then, when loop L1 ends, that is, when the processing for all the non-inherited folders including the files to be transferred is completed, the transfer destination determination unit 7 ends the operation.

[0070] The file transfer device 8 transfers files using information related to the transfer of files stored in the storage device 3. Here, the file transfer device 8 transfers file X and file Y of the source file server 1 to folder B of the destination file server 9. FIG. 11 is a diagram showing an example of the state after the transfer of files in the destination file server 9.

[0071] In the description so far, access rights have been used for folder inheritance and non-inheritance. For example, generally, compared with on-premises, the cloud often has less freedom in setting the inheritance and non-inheritance of folder access rights and the scope of the application destination of access rights. And there are many cases where file transfer from on-premises to the cloud is desired. Therefore, access rights are used for folder inheritance and non-inheritance. Specifically, inheritance of access rights means that the access rights in a folder are the same as those in the parent folder. Also, non-inheritance of access rights means that the access rights in a folder are different from those in the parent folder.

[0072] However, inheritance and non-inheritance in a folder are not limited to access rights. Inheritance and non-inheritance in a folder may be inheritance and non-inheritance of any attribute of the folder. For example, the attributes subject to inheritance and non-inheritance may be at least one of the storage attribute (archive attribute), compression attribute, and encryption attribute, or a combination of these and access rights. However, the attributes subject to inheritance and non-inheritance are not limited to the above, and may be attributes different from the above.

[0073] The transfer system 10 configured as described above, similar to the first embodiment, automates the parameter adjustment work, determines appropriate parameters, and reduces the burden on the user in determining the folder. The reason is that the destination determination device 4 of the transfer system 10 operates in the same manner as the destination determination device 4 in FIG. 1. That is, the destination determination device 4 of the transfer system 10 determines the folder to transfer the file between file servers with different folder attribute systems without requiring parameter adjustment work by the user or the like. As a result, the transfer system 10 automates the parameter adjustment work, determines the parameters based on appropriate criteria, and can determine the folder to transfer the file between file servers with different folder attribute systems. As a result, the destination determination device 4 reduces the burden on the user in determining the folder. Further, the file information acquisition device 2 of the transfer system 10 generates a learning file for creating a destination determination model. As a result, the transfer system 10 can reduce the man-hours of the user or the like and create a learning file. Further, the file transfer device 8 of the transfer system 10 transfers the file to the determined folder. As a result, the transfer system 10 can transfer the file to an appropriate folder between file servers with different folder attribute systems.

[0074] As a method for determining the folder to transfer the file, for example, non-hierarchical cluster analysis is used. Non-hierarchical cluster analysis is a method of gathering things with similar properties from a group in which things of different natures are mixed and creating clusters. Note that non-hierarchical cluster analysis is a method of classifying data into clusters that do not have a hierarchical structure, unlike hierarchical cluster analysis. When determining the folder to transfer the file between file servers, non-hierarchical cluster analysis analyzes the folder to transfer the file as a cluster using, for example, the properties of the file as a property. There are multiple methods as the method of non-hierarchical cluster analysis. For example, there is the k-means method as non-hierarchical cluster analysis. However, the method for determining the folder to transfer the file is not limited to non-hierarchical cluster analysis.

[0075] In order to improve the analysis accuracy in non-hierarchical cluster analysis and the like, it is necessary to appropriately set the parameters in the analysis. The parameters are, for example, the number of folders or the weights for the properties of the files used in the analysis, but are not limited thereto. And generally, users and the like often set the parameters. For example, the user adjusts the parameters based on the analysis results of the actual folders. Thus, the determination of the folder for transferring the current files requires the parameter adjustment work by users and the like, and requires a lot of man-hours. In addition, it is difficult to ensure objectivity in the determination of parameters based on human judgment.

[0076] The destination determination device 4 automates the parameter adjustment work and determines appropriate parameters. As a result, the destination determination device 4 reduces the load on the user in determining the folder. Further, since the destination determination device 4 uses the destination determination model generated by supervised machine learning, it can determine the destination more objectively with respect to manual adjustment.

[0077] The destination determination unit 7 may determine the destination folder using other values in addition to the harmonic mean of the confidence levels. For example, the destination determination unit 7 may determine the destination folder using the variance of the confidence levels. For example, the destination determination unit 7 may operate as follows. Similar to the description so far, the destination determination unit 7 calculates the harmonic mean of the confidence levels of each destination folder. Then, the destination determination unit 7 extracts the destination folders whose harmonic mean is equal to or greater than a predetermined threshold. Alternatively, the destination determination unit 7 extracts a predetermined number of destination folders from the ones with larger harmonic means. Alternatively, the destination determination unit 7 extracts the destination folders whose harmonic means are within a predetermined range from the maximum harmonic mean. When there is one extracted destination folder, the destination determination unit 7 determines that destination folder as the destination. When a plurality of destination folders are extracted, the destination determination unit 7 calculates the variance of the confidence levels of the files for each destination folder. Then, the destination determination unit 7 determines the destination folder with the minimum variance as the destination.

[0078] Note that the destination determination device 4 may operate in units of folders instead of files. For example, the teacher data creation unit 5 may create teacher data using the properties of the folder. Then, the learning unit 6 may generate a destination determination model for the folder. And the destination determination unit 7 may determine the destination for each folder using the destination determination model. When the destination determination model selects a plurality of folders, the destination determination unit 7 may obtain the confidence level for each folder to be transferred to the determined folder, and determine the folder with the maximum harmonic mean of the confidence levels as the destination folder. Further, the destination determination unit 7 may use the variance of the confidence levels of the folders. When the destination determination device 4 operates in units of folders, the file transfer device 8 may transfer files in units of folders.

[0079] Note that in some file systems, a folder may be implemented as a special file. For example, a folder may be implemented as a file for storing information for managing files. In this case, "determining the folder to transfer the folder" is the same operation as "determining the folder to transfer the file". That is, the operation of the destination determination device 4 in each embodiment may include "the operation of determining the folder to transfer the folder" as "the operation of determining the folder to transfer the file".

[0080] The transfer system 10 may not include the storage device 3. That is, each device included in the transfer system 10 may directly transmit and receive information. Also, the transfer system 10 may transfer files between the requested file servers in response to a user's request or the like, instead of the fixed source file server 1 and destination file server 9. That is, the transfer system 10 may change at least one of the source file server 1 and the destination file server 9. Also, the file information acquisition device 2 may use information stored in a storage device (not shown).

[0081] FIG. 12 is a block diagram showing an example of the configuration of a transfer system 11, which is another configuration of the transfer system 10. The transfer system 11 includes a file information acquisition device 2, a transfer destination determination device 4, and a file transfer device 8. The transfer system 11 is connected to the file server serving as the requested transfer source and the file server serving as the transfer destination in response to a request from a user or the like. Then, the file information acquisition device 2, the transfer destination determination device 4, and the file transfer device 8 operate in the same manner as the transfer system 10 in FIG. 2, except for directly transmitting and receiving information. Specifically, each device operates as follows.

[0082] The file information acquisition device 2 generates a learning file and outputs it to the transfer destination determination device 4. For example, the file information acquisition device 2 creates a learning file using information stored in a storage device (not shown) and outputs the created learning file to the transfer destination determination device 4. The transfer destination determination device 4 acquires the learning file from the file information acquisition device 2. The teacher data creation unit 5 of the transfer destination determination device 4 creates teacher data using the learning file. The learning unit 6 generates a transfer destination determination model using the teacher data. The transfer destination determination unit 7 determines the transfer destination folder of the file server serving as the transfer destination for the files in the transfer source folder of the transfer source file server using the generated transfer destination determination model. Then, the transfer destination determination device 4 outputs the determined transfer destination folder to the file transfer device 8. The file transfer device 8 transfers the files in the transfer source folder of the transfer source file server to the transfer destination folder of the file server serving as the transfer destination determined by the transfer destination determination device 4.

[0083] The transfer system 11 configured in this way can achieve the same effects as the transfer system 10. That is, the transfer system 11 can generate a learning file for creating a transfer destination determination model. Further, the transfer system 11 automates the adjustment operation of the parameters of the transfer destination determination model using the learning file, determines appropriate parameters, and can determine the folder to transfer files between file servers with different folder attribute systems. Further, the transfer system 11 can transfer files to the determined folder.

[0084] <Hardware Configuration> Next, the hardware configuration of the transfer destination determination device 4 will be described. Each component of the transfer destination determination device 4 may be configured by a hardware circuit. Alternatively, in the transfer destination determination device 4, each component may be configured using a plurality of devices connected via a network. For example, the transfer destination determination device 4 may be configured using cloud computing. Alternatively, in the transfer destination determination device 4, a plurality of components may be configured by one piece of hardware.

[0085] Alternatively, the transfer destination determination device 4 may be realized as a computer device including a CPU (Central Processing Unit), a ROM (Read Only Memory), and a RAM (Random Access Memory). In addition to the above configuration, the transfer destination determination device 4 may also be realized as a computer device further including a network interface circuit (NIC: Network Interface Circuit).

[0086] FIG. 13 is a block diagram showing an example of the hardware configuration of the transfer destination determination device 4. The transfer destination determination device 4 includes a CPU 610, a ROM 620, a RAM 630, a storage device 640, and a NIC 650, and constitutes a computer device. The CPU 610 reads a program from at least one of the ROM 620 and the storage device 640. Then, based on the read program, the CPU 610 controls the RAM 630, the storage device 640, and the NIC 650. And the computer including the CPU 610 controls these configurations and realizes the functions of the teacher data creation unit 5, the learning unit 6, and the transfer destination determination unit 7 shown in FIGS. 1, 2, and 12.

[0087] When implementing each function, the CPU 610 may use at least one of the RAM 630 and the storage device 640 as a temporary storage medium for programs and data. Further, the CPU 610 may read a program included in a recording medium 690 storing a computer-readable program using a recording medium reading device (not shown). Alternatively, the CPU 610 may receive a program from an external device (not shown) via the NIC 650, store it in at least one of the RAM 630 and the storage device 640, and operate based on the stored program.

[0088] The ROM 620 stores programs and fixed data executed by the CPU 610. The ROM 620 is, for example, a P-ROM (Programmable-ROM) or a flash ROM. The RAM 630 temporarily stores at least one of the programs and data executed by the CPU 610. The RAM 630 is, for example, a D-RAM (Dynamic-RAM). The storage device 640 stores data and programs that the transfer destination determination device 4 stores long-term. Further, the storage device 640 may operate as a temporary storage device of the CPU 610. The storage device 640 is, for example, a hard disk device, a magneto-optical disk device, an SSD (Solid State Drive), or a disk array device.

[0089] The ROM 620 and the storage device 640 are non-transitory recording media. On the other hand, the RAM 630 is a transitory recording medium. And the CPU 610 is operable based on a program stored in at least one of the ROM 620, the storage device 640, and the RAM 630. That is, the CPU 610 is operable using at least one of the non-transitory recording medium and the transitory recording medium.

[0090] The NIC650 relays the data exchange with an external device (not shown) via a network. The NIC650 is, for example, a LAN (Local Area Network) card. Further, the NIC650 may use not only wired but also wireless. The transfer destination determination device 4 in FIG. 13 configured as such can obtain the same effects as the transfer destination determination devices 4 in FIGS. 1, 2, and 12. The reason is that the CPU 610 of the transfer destination determination device 4 in FIG. 13 can realize the same functions as the transfer destination determination devices 4 such as in FIG. 1 based on a program.

[0091] Note that at least one of the file information acquisition device 2 and the file transfer device 8 may be realized by using a computer device configured with the hardware shown in FIG. 13. Alternatively, at least two or all of the file information acquisition device 2, the transfer destination determination device 4, and the file transfer device 8 may be realized by using a computer device configured with the same hardware. Alternatively, any one of the file information acquisition device 2, the transfer destination determination device 4, or the file transfer device 8 may include the storage device 3. Alternatively, the learning device 41 or the transfer destination determination device 42 may be realized by using a computer device configured with the hardware shown in FIG. 13.

[0092] Some or all of the above embodiments may be described as follows in the appended claims, but are not limited thereto.

[0093] (Appended Claim 1) Teacher data creation means for creating teacher data using a learning file; Learning means for generating a transfer destination determination model that outputs information regarding a transfer destination folder of a transfer destination file server in response to an input of properties of files in a transfer source folder of a transfer source file server by supervised machine learning using the teacher data; A learning device including the above.

[0094] (Appended Claim 2) Destination determination means for determining a destination folder of a file in a source folder of a source file server on a destination file server that is a transfer destination of the file in the source folder of the source file server, using a destination determination model that outputs information on the destination folder of the destination file server A destination determination device including

[0095] (Appendix 3) Teacher data creation means for creating teacher data using learning files, and Learning means for generating a destination determination model that outputs information on a destination folder of a destination file server that is a transfer destination in response to an input of properties of a file in a source folder of a source file server, by supervised machine learning using the teacher data, and Destination determination means for determining a destination folder of a file in a source folder of a source file server on a destination file server that is a transfer destination of the file in the source folder of the source file server, using the destination determination model A destination determination device including

[0096] (Appendix 4) When the destination determination model selects a plurality of destination folders, the destination determination means determines a destination folder to which the file in the source folder is transferred, based on the confidence level of the file in the source folder for each of the source folders selected by the destination determination model The destination determination device according to Appendix 3.

[0097] (Appendix 5) The destination determination means determines the destination folder based on the harmonic mean of the confidence levels The destination determination device according to Appendix 4.

[0098] (Appendix 6) The destination determination means determines the destination folder using the variance of the confidence levels The destination determination device according to Appendix 5.

[0099] (Appendix 7) The teacher data creation means creates teacher data using the properties of the files included in the learning file. The transfer destination determination device according to any one of Appendices 3 to 6.

[0100] (Appendix 8) The teacher data creation means extracts the features of the file based on the properties of the files included in the learning file, and creates teacher data using the extracted features. The transfer destination determination device according to Appendix 7.

[0101] (Appendix 9) The transfer destination determination device according to any one of Appendices 3 to 8, A file information acquisition device that generates a learning file, A file transfer device that transfers the files in the source folder of the source file server to the destination folder of the destination file server determined by the destination determination device Including a transfer system.

[0102] (Appendix 10) Create teacher data using the learning file, Generate a transfer destination determination model that outputs information regarding the destination folder of the destination file server that is the transfer destination in response to the input of the properties of the files in the source folder of the source file server by teacher-based machine learning that uses the teacher data, Determine the destination folder of the destination file server that is the transfer destination of the files in the source folder of the source file server using the transfer destination determination model. Transfer destination determination method.

[0103] (Appendix 11) The transfer destination determination device executes the transfer destination determination method according to Appendix 10, The file information acquisition device generates a learning file, The file transfer device transfers the files in the source folder of the source file server to the destination folder of the destination file server determined by the destination determination device. Transfer method.

[0104] (Appendix 12) The process of creating teacher data using a learning file, and the process of generating a destination determination model that outputs information regarding a destination folder of a destination file server according to an input of properties of files in a source folder of a source file server by supervised machine learning using the teacher data, and the process of determining a destination folder of a destination file server that is the transfer destination of a file in a source folder of a source file server using the destination determination model A recording medium that records a program for causing a computer to execute the above.

[0105] Although the present invention has been described with reference to the embodiments above, the present invention is not limited to the above embodiments. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.

Industrial Applicability

[0106] The present invention can be used in devices used in the field of information transfer between computers, the field of data communication, and the field of file transfer between on-premises and cloud, etc. Alternatively, the present invention can be used in devices that manage contents such as files and folders in devices such as shared storage.

Explanation of Signs

[0107] 1 Source file server 2 File information acquisition device 3 Storage device 4 Destination determination device 5 Teacher data creation unit 6 Learning unit 7 Destination determination unit 8 File transfer device 9 Destination file server 10 Transfer system 41 Learning device 42 Destination determination device 610 CPU 620 ROM 630 RAM 640 Memory device 650 NIC 690 Recording medium

Claims

1. Teacher data creation means for creating teacher data including a correct label which is information indicating a correct folder to be the transfer destination determined using the owner and access rights in the properties of the files included in the learning file; Learning means for generating a transfer destination determination model that outputs information regarding a transfer destination folder of a transfer destination file server to be the transfer destination in response to an input of the properties of the files in the transfer source folder of the transfer source file server by machine learning with a teacher using the teacher data; Transfer destination determination means for determining the transfer destination folder of the transfer destination file server to be the transfer destination of the files in the transfer source folder of the transfer source file server using the transfer destination determination model; comprising; when the transfer destination determination model selects a plurality of the transfer destination folders, the transfer destination determination means determines the transfer destination folder to transfer the files in the transfer source folder based on the harmonic mean of the confidence levels of each of the plurality of files in the transfer source folder with respect to each of the transfer destination folders selected by the transfer destination determination model; A transfer destination determination device.

2. The transfer destination determination means determines the transfer destination folder using the variance of the confidence levels of the plurality of files for each of the transfer destination folders. The transfer destination determination device according to Claim 1.

3. The transfer destination determination device according to Claim 1 or 2; A file information acquisition device for generating the learning file; A file transfer device for transferring the files in the transfer source folder of the transfer source file server to the transfer destination folder of the transfer destination file server determined by the transfer destination determination device; A transfer system comprising.

4. The transfer destination determination device creates teacher data including a correct label which is information indicating a correct folder to be the transfer destination determined using the owner and access rights in the properties of the files included in the learning file; generates a transfer destination determination model that outputs information regarding a transfer destination folder of a transfer destination file server to be the transfer destination in response to an input of the properties of the files in the transfer source folder of the transfer source file server by machine learning with a teacher using the teacher data; determines the transfer destination folder of the transfer destination file server to be the transfer destination of the files in the transfer source folder of the transfer source file server using the transfer destination determination model; When the transfer destination determination model selects a plurality of the transfer destination folders, the determination of the transfer destination folder is based on the harmonic mean of the confidence levels of each of the plurality of files in the source folder with respect to each of the transfer destination folders selected by the transfer destination determination model, and determines the transfer destination folder to which the files in the source folder are to be transferred. Transfer destination determination method.

5. A transfer destination determination device executes the transfer destination determination method according to claim 4. A file information acquisition device generates the learning file. A file transfer device transfers the files in the source folder of the source file server to the transfer destination folder of the transfer destination file server determined by the transfer destination determination device. Transfer method.

6. A process of creating teacher data including a correct label which is information indicating a correct folder serving as a transfer destination determined using the owner and access rights in the properties of the files included in the learning file. A process of generating a transfer destination determination model that outputs information regarding the transfer destination folder of the transfer destination file server serving as the transfer destination in response to the input of the properties of the files in the source folder of the source file server by supervised machine learning using the teacher data. A process of determining the transfer destination folder of the transfer destination file server serving as the transfer destination of the files in the source folder of the source file server using the transfer destination determination model. Causes a computer to execute. When the transfer destination determination model selects a plurality of the transfer destination folders, the process of determining the transfer destination folder is based on the harmonic mean of the confidence levels of each of the plurality of files in the source folder with respect to each of the transfer destination folders selected by the transfer destination determination model, and determines the transfer destination folder to which the files in the source folder are to be transferred. A program for causing execution.

Images