JP2020064581A - Information processing system - Google Patents

Abstract

To provide an information processing system capable of performing coordination between systems with simple and flexible methods, while taking advantage of current systems.SOLUTION: An information processing system includes a first reading part, a second reading part, and a learning part. The first reading part reads out first data from a first database for storing a plurality of the first data. The second reading part reads out second data from a second database for storing a plurality of the second data. The learning part performs machine learning by making: the first data of the first database and a first identifier as one of a plurality of identifiers serve as learning data; and the second data of the second database and a second identifier as one of the plurality of identifiers serve as the learning data.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an information processing system.

  In recent years, various work has been systematized, and labor saving of manual work is progressing. However, when the specifications of the systems are different, it is difficult for the systems to cooperate with each other. As a result, the user may need to individually operate various systems having different specifications when performing a certain task. For example, a user tried to extract data of another system (hereinafter, referred to as “data C”) based on data extracted from two systems (hereinafter, referred to as “data A, B”). In this case, due to the difference in specifications between the systems, the data C cannot be automatically extracted based on the data A and B, and the user may find the data C manually. In this case, manual data search is necessary, and it cannot be said that the systemization contributes to labor saving in manual work.

  Therefore, it is conceivable to replace existing systems having different specifications with systems having the same specifications so that the systems can cooperate with each other. Further, for example, as described in Patent Document 1, it is conceivable to provide a format correspondence table that defines a linkage relationship between data of each system.

International publication WO1015 / 016011

  However, to replace the system, it is necessary to redesign the format and association of data with overlapping meanings, data of different formats, and data of different sizes, perform data conversion, and input data to the new system. . Further, when the system is replaced or a format correspondence table is provided, it is not possible to change the data format or the data association in each database during the operation. Therefore, it is desirable to provide an information processing system capable of linking the systems by a simple and flexible method while making the most of the current system.

  An information processing system according to an embodiment of the present invention includes a first reading unit, a second reading unit, and a learning unit. The first reading unit reads the first data from a first database that stores a plurality of first data. The second reading unit reads the second data from a second database that stores a plurality of second data. The learning unit performs machine learning by using the first data read from the first database (first data of the first database) and the first identifier, which is one of the plurality of identifiers, as learning data. The learning unit further performs machine learning using the second data (second data of the second database) read from the second database and the second identifier, which is one of the plurality of identifiers, as learning data. .

  In the information processing system according to the embodiment of the present invention, machine learning is performed using the first data in the first database and the first identifier, which is one of the plurality of identifiers, as learning data. Further, machine learning is performed using the second data of the second database and the second identifier, which is one of the plurality of identifiers, as learning data. As described above, in the information processing system according to the embodiment of the present invention, two machine learning processes using a common identifier as learning data are performed in cooperation with two databases. As a result, it is not necessary to redesign the data format and data association in each database in order to link the two databases. Moreover, it is possible to change the data format and the association of data in each database without disturbing the cooperation of the two databases during the operation.

  According to the information processing system according to the embodiment of the present invention, two machine learning processes using a common identifier as learning data are performed in cooperation with two databases. Therefore, it is possible to easily and easily use the existing system. Systems can be linked in a flexible manner.

It is a figure showing an example of the functional block of the information processing system concerning one embodiment of the present invention. It is a figure showing an example of the hardware constitutions of the information processing system of FIG. It is a figure showing an example of the text data memorize | stored in the text information database and integrated information database of FIG. It is a figure showing the structural example of the image classification learning program of FIG. It is a figure showing an example of the image data memorize | stored in the image information database and integrated information database of FIG. It is a figure showing an example of the image classification name list memorize | stored in the integrated information database of FIG. It is a figure showing an example of the image data for learning of FIG. It is a figure showing an example of the learning data input into the image classification learning program of FIG. It is a figure showing an example of the learning data input into the image classification learning program of FIG. It is a figure showing an example of the learning data input into the text classification learning program of FIG. It is a figure showing an example of the screen displayed on GUI of FIG. It is a figure showing an example of the screen displayed on GUI of FIG. It is a figure showing an example of the screen displayed on GUI of FIG. It is a figure showing an example of the screen displayed on GUI of FIG. It is a figure showing an example of the data acquisition procedure in the information processing system of FIG. It is a figure showing an example of the machine learning procedure in the information processing system of FIG. It is a figure showing an example of the machine learning procedure in the information processing system of FIG. It is a figure showing an example of the test procedure in the information processing system of FIG. It is a figure showing an example of the test procedure in the information processing system of FIG. It is a figure showing an example of the key image acquisition procedure in the information processing system of FIG. It is a figure showing an example of the screen displayed on GUI of FIG.

Hereinafter, embodiments for carrying out the present invention will be described in detail with reference to the drawings. The following description is one specific example of the present invention, and the present invention is not limited to the following embodiments. Further, the present invention is not limited to the arrangement, dimensions, dimensional ratios, etc. of the constituent elements shown in the drawings. The description will be given in the following order.
1. Embodiment 2. Modification

<1. Embodiment>
[Constitution]
An information processing system 100 according to an embodiment of the present invention will be described. FIG. 1 illustrates an example of functional blocks of the information processing system 100.

  The information processing system 100 is a device that enables cooperation between systems having different specifications. Here, the “specification” refers to, for example, the format or association of data. Further, “cooperation between systems” means, for example, associating data of one system with data of the other system. When the system specifications are different from each other, it is difficult to use the data of the other system. By interposing the information processing system 100 between a plurality of systems having different specifications, the information processing system 100 makes it possible to use the data of the partner system in the systems having different specifications.

  The information processing system 100 can communicate with, for example, two systems having different specifications (image information database 200, text information database 300), and two systems having different specifications (image information database 200, text information database). 300) is a device that enables cooperation between each other. “Coordination between two systems (image information database 200, text information database 300)” means, for example, data of one system (text information database 300) and data of the other system (image information database 200). Refers to associating.

  The information processing system 100 includes, for example, GUI (Graphical User Interface) 110, image data IF (Interface) 120, image classification learning program 130, GUI (Graphical User Interface) 140, text data IF (Interface) 150, text classification learning program. 160, an integrated information database 170 and an information processing section 180.

  FIG. 2 shows an example of the hardware configuration of the information processing system 100. The information processing system 100 is connected to two systems (information processing systems 2000 and 3000) via a network 4000.

(Network 4000)
The network 4000 is, for example, a network that performs communication using a communication protocol (TCP / IP) that is standardly used on the Internet. The network 4000 may be, for example, a secure network that communicates using a communication protocol unique to the network. The network 4000 is, for example, the Internet, an intranet, or a local area network. The connection between the network 4000 and the information processing system 100, the information processing system 2000, or the information processing system 3000 may be, for example, a wired LAN (Local Area Network) such as Ethernet, or a wireless LAN such as Wi-Fi. It may be a mobile phone line or the like.

(Information processing system 2000)
The information processing system 2000 is configured to include, for example, a CPU (Central Processing Unit) 2100, a memory 2200, and a network IF (Interface) 2300. The CPU 2100 executes, for example, a web server program stored in the memory 2200, an operating system, or the like. The network IF 2300 is a communication interface for communicating with the information processing system 100 via the network 4000. The memory 2200 stores a program (for example, a web server program or an operating system) executed by the CPU 2100. The memory 2200 includes a RAM (Random Access Memory), a ROM (Read Only Memory), an auxiliary storage device (hard disk, etc.), and the like. The memory 2200 is configured to include the image information database 200.

(Information processing system 3000)
The information processing system 3000 is configured to include, for example, a CPU 3100, a memory 3200, and a network IF 3300. The CPU 3100 executes, for example, a web server program stored in the memory 3200, an operating system, or the like. The network IF 3300 is a communication interface for communicating with the information processing system 100 via the network 4000. The memory 3200 stores a program (for example, a web server program or an operating system) executed by the CPU 3100. The memory 3200 includes a RAM, a ROM, an auxiliary storage device (hard disk, etc.), and the like. The memory 3200 is configured to include the text information database 300.

(Information processing system 100)
The information processing system 100 is configured to include, for example, a CPU 1100, a memory 1200, a network IF 1300, an input device 1400, and an output device 1500. The CPU 1100 executes, for example, a web browser program stored in the memory 1200 and an operating system. The network IF 1300 is a communication interface for communicating with the information processing systems 2000 and 3000 via the network 4000. The memory 1200 stores a program executed by the CPU 1100 (for example, a web browser program or an operating system). The memory 1200 includes a RAM, a ROM, an auxiliary storage device (hard disk, etc.), and the like. The input device 1400 is composed of a keyboard, a touch panel, and the like. The output device 1500 includes a display and the like.

  The memory 1200 includes an integrated information database 170, an image classification learning program 130, and a text classification learning program 160. The GUI 110, the image data IF 120, the GUI 140, and the text data IF 150 are configured by the CPU 1100, the network IF 1300, the input device 1400, and the output device 1500.

  Next, details of each functional block will be described.

(Text information database 300)
The text information database 300 stores a plurality of text data 10. The text data 10 is, for example, surgery data composed of a plurality of column data 11 as shown in FIG. The column data 11 is composed of a tag 12 and concrete contents 13 corresponding to the tag 12. Examples of the tag 12 include a patient name, a patient ID (Identification), a doctor in charge, a disease name, an operation start date, an operation start time, an operation eye, or an operation method. Examples of the content 13 of the patient name in the tag 12 include the name of the patient to be operated on (eg, Taro Yamada). An example of the content 13 of the patient ID of the tag 12 is an identifier (for example, an identification number such as 012345) given to the patient undergoing the surgery. Examples of the content 13 of the doctor in charge of the tag 12 include the name of the doctor who performs the surgery (for example, doctor A). Examples of the content 13 of the disease name in the tag 12 include the name of the disease that caused the operation (for example, cataract). Examples of the content 13 of the tag 12 for the start of surgery include the date on which the surgery is performed (for example, September 10, 2018). Examples of the content 13 of the operation start time of the tag 12 include the time when the operation is performed (for example, 10:30 am). The content 13 of the tag 12 as the surgical eye includes the position (for example, right (R)) of the eye on which the surgery is performed among the pair of eyes. An example of the content 13 of the surgical operation method using the tag 12 is the name of the operation (for example, PEA + IOL).

  In the text data 10, the designated one or a plurality of column data 11 constitutes the column data set 11s. That is, it can be said that the text data 10 includes the column data set 11s, and the text information database 300 stores a plurality of column data sets 11s (second data). The column data set 11s may be designated in advance or may be designated by the user. The column data set 11s is composed of one or a plurality of column data 11. Hereinafter, the column data 11 included in the column data set 11s will be referred to as designated column data 11-1 for convenience. The designated column data 11-1 is the column data 11 useful for classifying the image data 20. In the column data set 11s, for example, the tag 12 includes patient ID column data 11, the tag 12 includes surgical eye column data 11, and the tag 12 includes surgical technique column data 11. If the image classification learning program 130 is composed of a plurality of image classification learning programs 131 and 132 set for each doctor as shown in FIG. 4, for example, the column data set 11s is The tag 12 may include the column data 11 of the doctor in charge. Below, the column data 11 in which the tag 12 is the doctor in charge is referred to as customized column data 11c.

(Image information database 200)
The image information database 200 stores a plurality of image data 20 (first data). The plurality of image data 20 stored in the image information database 200 includes, for example, a plurality of learning image data 21, a plurality of test image data 22, and a plurality of image data 23, as shown in FIG. . The learning image data 21 is sample data used in the later-described machine learning. The test image data 22 is sample data used in a test for confirming the results of machine learning described later. The image data 23 is data obtained by photographing a patient undergoing surgery during an examination.

(Integrated information database 170)
The integrated information database 170 is a database used to link two systems (image information database 200 and text information database 300) with each other. The integrated information database 170 stores, for example, a plurality of image data 20 read from the image information database 200 and a plurality of text data 10 or column data sets 11s read from the text information database 300. The integrated information database 170 further stores the image data 20 and the feature amount data 24 obtained by processing the image data 20 by a predetermined algorithm. The feature amount data 24 is, for example, data representing features of organs (for example, eyes, heart, lungs, liver, etc.) included in the image data 20, and machine learning in the image classification learning model 130 and key image acquisition described later. This is the data used in the case.

  The image classification name list 30 is further stored in the integrated information database 170. The image classification name list 30 includes a plurality of image classification names 31, as shown in FIG. 6, for example. The image classification name 31 is an identifier for distinguishing each image data 20 by classification when the plurality of image data 20 are classified from a predetermined viewpoint (for example, an inspection device used for imaging, an operating eye, etc.). . The image classification name 31 is configured to include, for example, a name for classifying the type of each image data 20 or a character string or a numerical sequence corresponding to the name. The integrated information database 170 further stores learning results obtained by machine learning in the image classification learning program 130 and the text classification learning program 160.

  FIG. 7 shows an example of the learning image data 21. In FIG. 7, “slit (light right)”, “slit (dark left)”, “slit (blue left)”, “optos (left)”, and “specular (right)” are assigned as the image classification name 31. A plurality of learning image data 21 to be obtained is illustrated. Here, the term “slit” refers to image data obtained by photographing a patient's eye with a slit lamp microscope. Optos refers to image data obtained by photographing with a fundus examination device manufactured by Optos. Specular refers to image data obtained by photographing with a corneal endothelial cell test device. “Bright” refers to image data obtained by shooting in a bright state. “Dark” refers to image data obtained by shooting in a dark state. “Blue” refers to image data obtained by shooting with blue light applied to the eye. “Right” refers to image data obtained by photographing the right eye. "Left" refers to image data obtained by photographing the left eye.

(Image data IF120)
The image data IF 120 is an interface for communicating between the information processing system 100 and the image information database 200. The image data IF 120 reads the plurality of image data 20 and the like from the image information database 200 and outputs them to the GUI 110 and the integrated information database 170, for example, under the control of the GUI 110.

(Text data IF150)
The text data IF 150 is an interface for communicating between the information processing system 100 and the text information database 300. The text data IF 150 reads a plurality of text data 10 (or the column data set 11s) and the like from the text information database 300 and outputs them to the GUI 140 and the integrated information database 170 under the control of the GUI 140, for example.

(Text classification learning program 160)
The text classification learning program 160 is loaded into the information processing unit 180 to machine-learn the input data as learning data and store the learning result obtained thereby in the integrated information database 170. The input data input to the text classification learning program 160 is, for example, a set of column data set 11s and image classification name 31 as shown in FIG. 8C.

  The information processing unit 180 loaded with the text classification learning program 160 creates a text classification model in the machine learning process. Specifically, in the machine learning process, the information processing unit 180 loaded with the text classification learning program 160, the text data 10 (or the column data set 11s) read from the text information database 300, and a plurality of image classifications. Machine learning is performed using the image classification name 31 (second identifier), which is one of the names 31 (identifiers), as learning data. As a result, the information processing unit 180 generates information about the correspondence between the text data 10 (or the column data set 11s) and the image classification name 31 as a learning result. In the key image acquisition process described later, when the text data 10 (or the column data set 11s) is input, the information processing unit 180 loaded with the text classification learning program 160 is input based on the learning result, for example. The most appropriate image classification name 31 (31x) for the extracted text data 10 (10x) (or column data set 11s) is extracted from the integrated information database 170. The information processing unit 180 loaded with the text classification learning program 160 compares, for example, the text data 10x (or column data set 11s) with various text data 10y (or column data set 11s), and as a result, various The image classification name 31x corresponding to the text data 10y (or column data set 11s) closest to the text data 10x (or column data set 11s) in the text data 10y (or column data set 11s) is extracted from the integrated information database 170. To do.

(Image classification learning program 130)
The image classification learning program 130 is loaded into the information processing unit 180, machine-learns the input data as learning data, and stores the learning result obtained thereby in the integrated information database 170. The input data input to the image classification learning program 130 is, for example, as shown in FIG. 8A, a set of learning image data 21 and an image classification name 31. The input data input to the image classification learning program 130 may be, for example, a set of feature amount data 24 and an image classification name 31 as shown in FIG. 8B. The information processing unit 180 loaded with the image classification learning program 130 performs machine learning without using the text even if text such as a file name or a meta tag is attached to the image data 20.

  The information processing unit 180 loaded with the image classification learning program 130 creates an image classification model in the machine learning process. Specifically, in the machine learning process, the information processing unit 180 loaded with the image classification learning program 130 selects the image data 20 read from the image information database 200 and the plurality of image classification names 31 (identifiers). Machine learning is performed using the image classification name 31 (first identifier) that is one of the above as learning data. As a result, the information processing unit 180 generates information about the correspondence between the image data 20 and the image classification name 31 as a learning result. When the image classification name 31x is input in the key image acquisition process described later, the information processing unit 180 loaded with the image classification learning program 130, for example, adds the input image classification name 31x based on the learning result. The most appropriate image data 20 (key image 20x) is extracted from the integrated information database 170 from the corresponding many image data 20.

(GUI110)
The GUI 110 provides a screen for a user to input data to the information processing system 100 or to present data to the user from the information processing system 100. The GUI 110 provides a screen 111 for creating an image classification model, as shown in FIG. 9, for example. On the screen 111, for example, an image classification name list 30 necessary for executing machine learning and a test for confirming the results of machine learning is displayed. The screen 111 further displays, for example, learning image data 21 that has been machine-learned and test image data 22 that has been tested. On the screen 111, for example, a button 113 for adding or deleting the image classification name 31, a button 114 for adding or deleting the learning image data 21, and a button 115 for executing machine learning. A button 116 for adding or deleting the test image data 22 and a button 117 for executing the test are displayed.

  The GUI 110 further provides a screen 112 for setting the image classification model 40, for example, as shown in FIG. The image classification model 40 refers to a model regarding a hierarchical structure of a plurality of image classification names 31 included in the image classification name list 30. On the screen 112, for example, a model in which all registered image classification names 31 are listed without a hierarchy is displayed in the list window 118. The screen 112 further displays a button 119 for setting the hierarchical structure of the image classification model 40, for example. Note that FIG. 10 illustrates an example in which the image classification model 40 is configured by three image classification models 40A, 40B, and 40C, and two image classification models 40B and 40C are set in the lower layer of the image classification model 40A. ing. At this time, for example, the uppermost layer image classification model 40A is configured with six image classification names 31A-1 to 31A-6, and the lower layer image classification model 40B is connected to the image classification name 31A-2 to obtain the image classification model. 130B is composed of two image classification names 31B-1 and 31B-2, an image classification model 40C is connected to the image classification name 31A-5, and the image classification model 40C is divided into three image classification names 31C-1 to 31C-1. 31C-3. When the image classification model 40 has such a hierarchical structure, in the list window 118, the image classification name 31A-2 is replaced with the image classification names 31B-1 and 31B-2 in the image classification model 40A, and the image classification name is changed. A display in which 31A-5 is replaced with the image classification names 31C-1, 31C-2, 31C-3 is displayed.

(GUI140)
The GUI 140 provides a screen for a user to input data to the information processing system 100 or to present data to the user from the information processing system 100. The GUI 140 provides a screen 141 for creating the tag set 12s as shown in FIG. 11, for example. On the screen 141, for example, all the designated tags 12 (designated tags 12-1, 12-2, 12-3) are displayed in a list. On the screen 141, for example, a button 143 for adding the tag 12 to the tag set 12s and a button for selecting the image classification model 40 associated with the tag set 12s from the plurality of image classification models 40. 144 is displayed. Note that in FIG. 11, the image classification model 40 selected as the image classification model 40 associated with the tag set 12s (a model in which all the image classification names 31 are listed without hierarchy) is displayed on the screen 141. It is illustrated.

  The GUI 140 provides a screen 142 for creating a text classification model, for example, as shown in FIG. On the screen 142, for example, a list of a plurality of text data 10 (10A) necessary to execute machine learning, and a plurality of text data necessary to execute a test for confirming the results of machine learning ( A list of 10B) is displayed. Further, on the screen 142, for example, the set tag set 12s and the image classification name list 30 including the set image classification name 31 are displayed. The screen 142 further displays, for example, a button 145 for executing machine learning and a button 146 for executing a test.

[motion]
Next, the operation of the information processing system 100 according to this embodiment will be described.

(Image data acquisition)
13 and 14 show an example of a data acquisition procedure in the information processing system 100. The user acquires the data (image data 20) in the image information database 200. Specifically, the user instructs the GUI 110 to display a screen for acquiring the data (image data 20) in the image information database 200. As a result, image acquisition is started. Then, the GUI 110 displays a screen for acquiring the data (image data 20) in the image information database 200. Next, the user specifies a condition for acquiring the image data 20 on the GUI 110. Then, the information processing unit 180 outputs an image acquisition request according to the condition designated by the GUI 110 to the image data IF 120. At this time, the image acquisition request includes, for example, the transmission destination of the acquired image (specifically, the integrated information database 170). The image data IF 120 outputs the input image acquisition request to the image information database 200. When the image acquisition request is input, the image information database 200 reads out the image data 20 that matches the conditions included in the input image acquisition request, and transmits the image data 20 to the image data IF 120. Upon receiving the image data 20 from the image information database 200, the image data IF 120 transmits the received image data 20 to the destination of the image included in the image acquisition request (specifically, the integrated information database 170). When receiving the image data 20 from the image data IF 120, the integrated information database 170 stores the received image data 20. Thus, the acquisition of the image data 20 is completed.

(Acquire text data 10)
The user acquires the text data 10 (or the column data set 11s) of the text information database 300. Specifically, the user instructs the GUI 140 to display a screen for acquiring the text data 10 (or the column data set 11s) of the text information database 300. This starts the text acquisition. Then, the GUI 140 displays a screen for acquiring the text data 10 (or the column data set 11s) of the text information database 300. Next, the user specifies, on the GUI 140, a condition for acquiring the text data 10 (or the column data set 11s). Then, the information processing section 180 outputs a text acquisition request according to the condition designated by the GUI 140 to the text data IF 150. At this time, the text acquisition request includes, for example, the transmission destination (specifically, the integrated information database 170) of the acquired text data 10 (or the column data set 11s). The text data IF 150 outputs the input text acquisition request to the text information database 300. When the text acquisition request is input, the text information database 300 reads the text data 10 (or the column data set 11s) that matches the conditions included in the input text acquisition request, and transmits the text data 10 to the text data IF 150. When the text data IF 150 receives the text data 10 (or the column data set 11s) from the text information database 300, the text data 10 (or the column data set 11s) includes the text data 10 (or the column data set 11s) included in the text acquisition request. The column data set 11s) is transmitted to the transmission destination (specifically, the integrated information database 170). Upon receiving the text data 10 (or the column data set 11s) from the text data IF 150, the integrated information database 170 stores the received text data 10 (or the column data set 11s). In this way, the acquisition of the text data 10 (or the column data set 11s) of the text information database 300 is completed.

(Machine learning)
FIG. 15 shows an example of a machine learning procedure for the image data 20. First, the user instructs the GUI 110 to display a screen 112 for setting the image classification model 40. Then, the GUI 110 displays a screen 112 for setting the image classification model 40. Next, the user sets the hierarchical structure of the image classification model 40 as needed. The user configures the image classification model 40A of the uppermost layer with six image classification names 31A-1 to 31A-6, connects the image classification model 40B of the lower layer to the image classification name 31A-2, and 40B is composed of two image classification names 31B-1 and 31B-2, an image classification model 31C-5 is connected to an image classification model 40C in the lower layer, and the image classification model 40C is divided into three image classification names 31C-1 to 31C-1. 31C-3. The information processing unit 180 transmits the setting information about the hierarchical structure of the image classification model 40, which is input via the GUI 110, to the integrated information database 170. The integrated information database 170 stores the received setting information.

  Next, the user instructs the GUI 110 to display the screen 111 of the image classification model creation UI. Then, the GUI 110 displays the screen 111 of the image classification model creation UI. Next, the user instructs the GUI 110 to read out the plurality of image data 20 by pressing the button 114 on the screen 111, for example. When the information processing unit 180 receives a read instruction (image acquisition request) for the plurality of image data 20 via the GUI 110, the information processing unit 180 starts machine learning for the image data 20. The information processing unit 180 outputs the image acquisition request to the integrated information database 170. At this time, the image acquisition request includes, for example, the transmission destination of the acquired image (specifically, the GUI 110). When the image acquisition request is input, the integrated information database 170 reads the plurality of image data 20 and sends the image data 20 to the GUI 110. Upon receiving the plurality of image data 20, the GUI 110 displays the received plurality of image data 20 on the screen 111. The user selects one suitable for machine learning from the plurality of image data 20 displayed on the screen 111. Subsequently, the user classifies the image classification name 31 suitable for the selected image data 20 by pressing the button 113 on the screen 111 and inputting the image classification name 31 suitable for the selected image data 20. Add to name list 30. When the image classification name 31 suitable for the selected image data 20 is already displayed on the screen 111, the user selects the image classification name 31 suitable for the selected image data 20. At this time, the information processing unit 180 generates the data of the screen 111 so that the selected image data 20 and the image classification name 31 suitable for the input or selected image data 20 are paired and transferred to the GUI 110. To do. To display. The GUI 110 displays the screen 111 based on the data of the screen 111 from the information processing unit 180.

  Next, the user presses the button 115 on the screen 111. Then, the GUI 110 first sends the image classification name list 30 to the transmission destination associated with the button 115 (specifically, the integrated information database 170). The integrated information database 170 stores the received image classification name list 30. The GUI 110 further uses, as learning data, the image data 20 (learning image data 21) and the image classification name 31 suitable for the image data 20 (learning image data 21), as a destination associated with the button 115 ( Specifically, it is sent to the integrated information database 170 and the information processing unit 180). The integrated information database 170 stores the received set of image data 20 (learning image data 21) and the image classification name 31 as learning data. At this time, the information processing unit 180 converts the image data 20 into the feature amount data 24, and uses the feature amount data 24 obtained by the conversion and the image classification name 31 suitable for the image data 20 before the conversion as learning data. Machine learning is performed and the learning result is sent to the integrated information database 170. The integrated information database 170 stores the received learning result. In this way, the machine learning for the image data 20 ends.

  Next, the user instructs the GUI 140 to display the screen 141 of the tag set creation UI. Then, the GUI 140 displays the screen 141 of the tag set creation UI. Next, the user instructs the GUI 140 to create the tag set 12s by pressing the button 143 on the screen 141, for example. As a result, the creation of the tag set 12s is started. When the information processing unit 180 receives the acquisition request for the text data 10 via the GUI 140, the information processing unit 180 outputs the received acquisition request for the text data 10 to the integrated information database 170. Here, the acquisition request for the text data 10 includes, for example, the transmission destination (specifically, the GUI 140) of the acquired text data 10. When the acquisition request for the text data 10 is input, the integrated information database 170 reads the text data 10 and sends it to the destination (specifically, the GUI 140) included in the acquisition request for the text data 10. Upon receiving the text data 10, the GUI 140 displays the received text data 10 on the screen 141. The user selects one or more tags 12 from the displayed text data 10. Then, the information processing section 180 generates the tag set 12s having the one or more tags 12 selected by the user as the designated tags (for example, 12-1, 12-2, 12-3), and the integrated information database 170. Send to. The integrated information database 170 stores the received tag set 12s. In this way, the setting of the tag set 12s is completed.

  The user further instructs the GUI 140 to select the image classification model 40 by pressing the button 144 on the screen 141, for example. As a result, selection of the image classification model 40 is started. When the information processing unit 180 receives the selection instruction of the image classification model 40 via the GUI 140, the information processing unit 180 reads the setting information about the hierarchical structure of the image classification model 40 from the integrated information database 170 and sends it to the GUI 140. Upon receiving the setting information about the hierarchical structure of the image classification model 40 from the information processing unit 180, the GUI 140 displays the received setting information about the hierarchical structure of the image classification model 40 on the screen 141. The user selects one structure from the setting information about the hierarchical structure of the displayed image classification model 40. When the structure of the image classification model 40 is selected by the user, the GUI 140 displays the image classification model 40 of the selected structure. The information processing unit 180 sends information about the image classification model 40 selected by the user to the integrated information database 170. The integrated information database 170 stores the received selection information about the image classification model 40. In this way, the selection of the image classification model 40 is completed.

  FIG. 16 illustrates an example of a machine learning procedure for the text data 10 (or the column data set 11s). After the machine learning on the image data 20 is completed, the user instructs the GUI 140 to display the screen 142 of the text classification model creation UI. As a result, machine learning on the text data 10 (or the column data set 11s) is started. When the information processing unit 180 receives a display instruction of the screen 142 of the text classification model creation UI via the GUI 140, the information processing unit 180 creates data of the screen 142 of the text classification model creation UI and sends it to the GUI 140. The GUI 140 displays the screen 142 based on the data of the screen 142 received from the information processing unit 180. The information processing unit 180 further requests the integrated information database 170 to acquire the plurality of learning text data 10A (or the column data set 11s). The acquisition request includes, for example, the transmission destinations (specifically, the GUI 140) of the plurality of read text data 10 (or the column data set 11s). When the integrated information database 170 receives the acquisition request for the plurality of learning text data 10A (or the column data set 11s), the integrated information database 170 reads the plurality of text data 10 (or the column data set 11s) and transmits the data included in the acquisition request. It is sent to the destination (specifically, GUI 140). When the GUI 140 receives the plurality of text data 10 (or the column data set 11s), the GUI 140 converts the received plurality of text data 10 (or the column data set 11s) into a plurality of learning text data 10A (or the column data set 11s). Is displayed on the screen 142. At this time, the information processing section 180 further requests the integrated information database 170 to acquire the image classification name list 30. The acquisition request includes, for example, the transmission destinations (specifically, the GUI 140) of the plurality of read text data 10 (or the column data set 11s). Upon receiving the acquisition request for the image classification name list 30, the integrated information database 170 reads the image classification name list 30 and sends it to the destination (specifically, the GUI 140) included in the acquisition request. Upon receiving the image classification name list 30, the GUI 140 displays the received image classification name list 30 on the screen 142. Next, the user, for example, selects one text data 10A (for example, text data 10A-1) (or column data) from the plurality of learning text data 10A (or column data set 11s) displayed on the screen 142. Select the set 11s) and press the button 145 on the screen 142. Then, the GUI 140 displays the selected text data 10A (for example, text data 10A-1) or the column data set 11s included in the selected text data 10A (for example, text data 10A-1) and the image classification name list. And are transmitted as learning data to the transmission destination (specifically, the integrated information database 170) associated with the button 145. The integrated information database 170 stores the received data as learning data. At this time, the GUI 140 selects the selected text data 10A (for example, text data 10A-1) or the column data set 11s included in the selected text data 10A (for example, text data 10A-1) and the image classification name. The list and the learning data are sent to the transmission destination (specifically, the information processing unit 180) associated with the button 145. Note that the GUI 140 may send the selected column data set 11s and the image classification name list to the integrated information database 170 as learning data. Furthermore, the GUI 140 may send the selected column data set 11s and the image classification name list as learning data to the information processing unit 180. The information processing unit 180 performs machine learning by using the received data as learning data and sends the learning result to the integrated information database 170. The integrated information database 170 stores the received learning result. In this way, the learning ends.

(test)
FIG. 17 shows an example of a test procedure for confirming the machine learning result of the image data 20. First, the user instructs the GUI 110 to display the screen 111 of the image classification model creation UI. When the information processing unit 180 receives a display instruction of the screen 111 of the image classification model creation UI via the GUI 110, the information processing unit 180 generates data of the screen 111 of the image classification model creation UI and sends it to the GUI 110. The GUI 110 displays the screen 111 based on the data of the screen 111 received from the information processing unit 180. The information processing unit 180 further outputs an acquisition request for the image classification name list 30 to the integrated information database 170. Here, the acquisition request of the image classification name list 30 includes, for example, the transmission destination (specifically, the GUI 110) of the read image classification name list 30. When the acquisition request for the image classification name list 30 is input, the integrated information database 170 reads the image classification name list 30 and sends it to the transmission destination (specifically, the GUI 110) included in the acquisition request for the image classification name list 30. . Upon receiving the image classification name list 30, the GUI 110 displays the received image classification name list 30 on the screen 111.

  Next, the user instructs the GUI 110 to acquire the plurality of image data 20 by pressing the button 116 on the screen 111, for example. As a result, the test using the image data 20 is started. When the information processing unit 180 receives an acquisition request for a plurality of image data 20 (image acquisition request) via the GUI 110, the information processing unit 180 outputs the received image acquisition request to the integrated information database 170. Here, the image acquisition request includes, for example, the transmission destination (specifically, the GUI 110) of the acquired plurality of image data 20. When the image acquisition request is input, the integrated information database 170 reads out the plurality of image data 20 and sends it to the destination (specifically, the GUI 110) included in the image acquisition request. Upon receiving the plurality of image data 20, the GUI 110 displays the received plurality of image data 20 on the screen 111. The user selects one suitable for the test image data 22 from the plurality of image data 20 displayed on the screen 111. Then, the GUI 110 displays the plurality of selected image data 20 as the test image data 22 on the screen 111.

  Next, the user presses the button 117 on the screen 111. Then, the GUI 110 sends the selected image data 20 (test image data 22) to the information processing unit 180. The information processing unit 180 estimates the image classification name 31 of the image data 20 (test image data 22) from the received image data 20 (test image data 22). At this time, the information processing unit 180, for example, converts the received image data 20 (test image data 22) into feature amount data 24, and based on the feature amount data 24 obtained thereby, the image data 20 (test The image classification name 31 of the use image data 22) is estimated.

  The GUI 110 transmits the image classification name 31 obtained by the estimation to the GUI 110. The GUI 110 displays the received image classification name 31 on the screen 111 as the image classification name 31 of the selected image data 20 (test image data 22). Note that the GUI 110 may display the selected image data 20 (test image data 22) at a position adjacent to a position where the same image classification name 31 as the received image classification name 31 is displayed.

  When the user determines that the image classification name 31 presented by the information processing unit 180 (image classification learning program 130) is not suitable as the image classification name 31 of the selected image data 20 (test image data 22). In addition, the user corrects the image classification name 31 presented by the information processing unit 180 (image classification learning program 130) to an appropriate image classification name 31. Note that when the GUI 110 displays the selected image data 20 (test image data 22) at a position adjacent to the position where the same image classification name 31 as the received image classification name 31 is displayed, May move the selected image data 20 (test image data 22) to a location adjacent to the location where the appropriate image classification name 31 is displayed.

  A location where the image classification name 31 is corrected to an appropriate image classification name 31, or the selected image data 20 (test image data 22) is adjacent to a location where the appropriate image classification name 31 is displayed. When moved to, the GUI 110 sends the selected image data 20 (test image data 22) and the appropriate image classification name 31 to the integrated information database 170 as learning data. The integrated information database 170 stores the received pair of image data 20 (test image data 22) and the image classification name 31 as learning data. At this time, the GUI 110 converts the selected image data 20 (test image data 22) into feature amount data 24, and the feature amount data 24 obtained by the conversion and the pre-conversion image data 20 (test image data). The image classification name 31 suitable for 22) is sent to the information processing unit 180 as learning data. The information processing unit 180 performs machine learning with the received pair of feature amount data 24 and the image classification name 31 as the first learning data, and sends the learning result to the integrated information database 170. The integrated information database 170 stores the received learning result.

  FIG. 18 shows an example of a test procedure for confirming the machine learning results for the text data 10 (or the column data set 11s). First, the user instructs the GU 1140 to display the screen 142 of the text classification model creation UI. When the information processing unit 180 receives a display instruction of the screen 142 of the text classification model creation UI via the GUI 140, the information processing unit 180 creates data of the screen 142 of the text classification model creation UI and sends it to the GUI 140. The GUI 140 displays the screen 142 based on the data of the screen 142 received from the information processing unit 180. The information processing unit 180 further outputs an acquisition request for the text data 10 (or the column data set 11s) to the integrated information database 170. Here, the acquisition request includes, for example, the transmission destination (specifically, the GUI 140) of the read text data 10 (or the column data set 11s). When the acquisition request for the text data 10 (or the column data set 11s) is input, the integrated information database 170 reads out the plurality of text data 10 (or the column data set 11s) and includes the text data 10 included in the acquisition request. (Or the column data set 11s) is sent to the destination (GUI 140). The information processing unit 180 further displays a list of the received plurality of text data 10 (or column data set 11s) on the screen 142 as a plurality of test text data 10B (or column data set 11s). The information processing unit 180 further outputs an acquisition request for the image classification name list 30 to the integrated information database 170. The acquisition request for the image classification name list 30 includes, for example, the transmission destination of the image classification name list 30 (specifically, the GUI 140). When the acquisition request for the image classification name list 30 is input, the integrated information database 170 reads the image classification name list 30 and sends it to the destination (specifically, the GUI 140) included in the acquisition request for the image classification name list 30. . Upon receiving the image classification name list 30, the GUI 140 displays a list of the received image classification name list 30 on the screen 142.

  Next, the user selects one text data 10B (or column data set 11s) from the plurality of text data 10B (or column data set 11s) displayed on the screen 142 and presses the button 146. Then, the GUI 140 transmits the column data set 11s in the selected text data 10B or the selected column data set 11s to the information processing unit 180. The information processing unit 180 estimates the most appropriate image classification name 31 for the received column data set 11s and transmits the image classification name 31 obtained thereby to the GUI 140. The GUI 140 displays the received image classification name 31 on the screen 142. When the image classification name list 30 is displayed on the screen and the image classification name 31 that is the same as the received image classification name 31 is included in the image classification name list 30, the GUI 140 displays The same image classification name 31 as the received image classification name 31 in the name list 30 may be highlighted.

  When the user determines that the image classification name 31 presented by the information processing unit 180 (text classification learning program 160) is not suitable as the image classification name 31 of the selected text data 10B (or column data set 11s). First, the user corrects the image classification name 31 presented by the text classification learning program 160 to an appropriate image classification name 31.

  When the image classification name 31 is corrected to the appropriate image classification name 31, the GUI 140 uses the selected text data 10B (or column data set 11s) and the appropriate image classification name 31 as learning data, and the integrated information database. Send to 170. The integrated information database 170 stores the received pair of text data 10B (or column data set 11s) and the image classification name 31 as learning data. At this time, the GUI 140 sends the column data set 11s of the selected text data 10B and the appropriate image classification name 31 as learning data to the information processing section 180 loaded with the text classification learning program 160. The information processing unit 180 loaded with the text classification learning program 160 performs machine learning using the received pair of column data sets 11s and the image classification name 31 as learning data, and sends the learning result to the integrated information database 170. The integrated information database 170 stores the received learning result.

(Key image acquisition)
FIG. 19 shows an example of a key image extraction procedure. The GUI 140 applies the learning result of the image classification learning program 130 and the text classification learning program 160 to the text data 10 (or the column data set 11s) read from the text information database 300 in the key image extraction process. Thus, the image data 20 related to the text data 10 (or the column data set 11s) is acquired.

  Specifically, first, the user instructs the GUI 140 to display the surgery list screen 147. When the information processing unit 180 receives the display instruction of the surgery list screen 147 via the GUI 140, the information processing unit 180 generates data of the surgery list screen 147 and sends the data to the GUI 140. The GUI 140 displays the screen 147 based on the data of the screen 147 received from the information processing unit 180. The information processing unit 180 further outputs an acquisition request for the text data 10 (or the column data set 11s) that matches a predetermined condition to the integrated information database 170. Here, the "text data 10 that matches a predetermined condition" refers to, for example, the text data 10 (or the column data set 11s) of the patient scheduled to undergo surgery today. The acquisition request for the text data 10 (or the column data set 11s) includes, for example, the transmission destination (specifically, the GUI 140, the information processing unit 180) of the read text data 10 (or the column data set 11s). .

  When an acquisition request for the text data 10 that matches a predetermined condition is input, the integrated information database 170 reads a plurality of text data 10 (or the column data set 11s) that matches the predetermined condition, and the text data 10 (or The column data set 11s) is sent to the destination (specifically, the GUI 140) included in the acquisition request. Upon receiving the plurality of text data 10 (or the column data set 11s), the GUI 140 associates the contents included in the received plurality of text data 10 (or the column data set 11s) with the display format of the surgery list and displays the screen 147. Is displayed in a list (see FIG. 20). At this time, the GUI 140, for example, displays the contents included in the plurality of received text data 10 (or the column data set 11s) in a list on the screen 147 in order of earliest operation start time.

  The integrated information database 170 further transmits a plurality of text data 10 (or column data set 11s) that matches a predetermined condition to the information processing unit 180. The information processing unit 180 estimates the most appropriate image classification name 31 for each of the received text data 10 (or column data set 11s) for each text data 10 (or column data set 11s) and obtains it. The most appropriate image data 20 (key image 20x) for a plurality of image classification names 31 is estimated. The information processing unit 180 reads the corresponding plurality of image data 20 (key image 20x) from the integrated information database 170 based on the estimation result. The information processing unit 180 transmits the plurality of read image data 20 (key image 20x) to the GUI 140. The GUI 140 displays the plurality of received image data 20 (key image 20x) on the screen 147. At this time, the GUI 140 allocates the plurality of received image data 20 (key images 20x) to the corresponding text data 10 (or column data set 11s) and displays them on the screen 147. In this way, the GUI 140 performs an operation in which all or part of the data included in the text data 10 (or the column data set 11s) is associated with the image data 20 related to the text data 10 (or the column data set 11s). A list is generated and displayed on the screen 147.

  The user determines that the image data 20 (key image 20x) presented by the information processing unit 180 (image classification learning program 130) is not suitable in relation to the corresponding text data 10 (or column data set 11s). In this case, the user corrects the image data 20 (key image 20x) presented by the image classification learning program 130 into appropriate image data 20. For example, the user presses the "wrong" button on the right side of the surgery list. As a result, the GUI 140 sends an image list acquisition request to the information processing unit 180. When the image list acquisition request is input from the GUI 140, the information processing unit 180 sends the image list acquisition request to the integrated information database 170. Here, the image list acquisition request includes, for example, the transmission destination of the read image data 20 (specifically, the GUI 140). When the integrated information database 170 receives the image list acquisition request, the integrated information database 170 reads out the plurality of image data 20 relating to the patient whose “wrong” button has been pressed, and transmits the image data 20 to the GUI 140. The GUI 140 displays the received plurality of image data 20 on the screen 147. The user selects appropriate image data 20 from the plurality of image data 20 displayed on the screen 147. The GUI 140 replaces the original image data 20 with the selected image data 20 and displays it on the screen 147.

  The GUI 140 further sends the image data 20 after replacement and the image classification name 31 corresponding to the image data 20 before replacement as learning data to the integrated information database 170. The integrated information database 170 stores the received pair of image data 20 (test image data 22) and the image classification name 31 as learning data. At this time, the GUI 140 sends the image data 20 after replacement and the image classification name 31 corresponding to the image data 20 before replacement as learning data to the information processing unit 180. The information processing unit 180 performs machine learning using the image data 20 after replacement and the image classification name 31 corresponding to the image data 20 before replacement as learning data, and sends the learning result to the integrated information database 170. The integrated information database 170 stores the received learning result. In this way, when the GUI 140 receives from the user a process for replacing the image data 20 included in the surgery list with another image data 20, the information processing unit 180 loaded with the image classification learning program 130 replaces the image data 20. Machine learning is performed using the subsequent image data 20 and the image classification name 31 corresponding to the image data 20 before replacement as learning data.

[effect]
Next, effects of the information processing system 100 according to the present embodiment will be described.

  In recent years, various work has been systematized, and labor saving of manual work is progressing. However, when the specifications of the systems are different, it is difficult for the systems to cooperate with each other. As a result, the user may need to individually operate various systems having different specifications when performing a certain task. For example, a user tried to extract data of another system (hereinafter, referred to as “data C”) based on data extracted from two systems (hereinafter, referred to as “data A, B”). In this case, due to the difference in specifications between the systems, the data C cannot be automatically extracted based on the data A and B, and the user may find the data C manually. In this case, manual data search is necessary, and it cannot be said that the systemization contributes to labor saving in manual work.

  Therefore, it is conceivable to replace existing systems having different specifications with systems having the same specifications so that the systems can cooperate with each other. Further, for example, as described in Patent Document 1, it is possible to provide a format correspondence table that defines a linkage relationship between data of each system.

  However, to replace the system, it is necessary to redesign the format and association of data with overlapping meanings, data of different formats, and data of different sizes, perform data conversion, and input data to the new system. . Further, when the system is replaced or a format correspondence table is provided, it is not possible to change the data format or the data association in each database during the operation.

  On the other hand, in the present embodiment, the image data 20 in the image information database 200 and the image classification name 31 that is one of the plurality of image classification names 31 that distinguish the type of each image data 20 are used as learning data. Machine learning is performed in the information processing unit 180 loaded with the image classification learning program 130. Further, the column data set 11s included in the text data 10 of the text information database 300, or the column data set 11s of the text information database 300 and the image classification name 31 that is one of the plurality of image classification names 31 is set. As learning data, machine learning is performed in the information processing unit 180 loaded with the text classification learning program 160.

  As described above, in the present embodiment, two machine learning processes using the common identifier (image classification name 31) as learning data are performed in cooperation with the two databases (image information database 200, text information database 300). Therefore, in order to link the two databases (the image information database 200 and the text information database 300), it is necessary to redesign the data format and the data association in each database (the image information database 200 and the text information database 300). Absent. Moreover, during the operation, the data formats and the data associations in the two databases (image information database 200, text information database 300) can be performed without hindering the cooperation of the two databases (image information database 200, text information database 300). You can also change the. Therefore, while utilizing the current system, it is possible to perform the cooperation between the systems by a simple and flexible method.

  Further, in the present embodiment, two learning results by machine learning are applied to the column data set 11s included in the text data 10 of the text information database 300 or the column data set 11s of the text information database 300. Thus, the image data 20 (key image data) related to the column data set 11s is acquired. Therefore, in order to link the two databases (the image information database 200 and the text information database 300), it is necessary to redesign the data format and the data association in each database (the image information database 200 and the text information database 300). Absent. Moreover, during the operation, the data formats and the data associations in the two databases (image information database 200, text information database 300) can be performed without hindering the cooperation of the two databases (image information database 200, text information database 300). You can also change the. Therefore, while utilizing the current system, it is possible to perform the cooperation between the systems by a simple and flexible method.

  Further, in the present embodiment, the column data set 11s included in the text data 10 of the text information database 300 or the column data set 11s of the text information database 300 is stored in the integrated information database 170. As a result, it is not necessary to frequently access an external database (image information database 200, text information database 300) when acquiring the text data 10 or the column data set 11s. You can

  Further, in the present embodiment, the image data 20 and the text data 10 or the column data set 11s are associated with each other using the identifier of the image classification name 31. Therefore, in order to link the two databases (the image information database 200 and the text information database 300), it is necessary to redesign the data format and the data association in each database (the image information database 200 and the text information database 300). Absent. Therefore, while utilizing the current system, it is possible to perform the cooperation between the systems by a simple and flexible method.

  Further, in the present embodiment, the image classification name 31 used as an identifier is configured to include a name for classifying the type of the image data 20 or a character string or number sequence corresponding thereto. Therefore, in order to link the two databases (the image information database 200 and the text information database 300), it is necessary to redesign the data format and the data association in each database (the image information database 200 and the text information database 300). Absent. Therefore, while utilizing the current system, it is possible to perform the cooperation between the systems by a simple and flexible method.

  Further, in the present embodiment, even if a text (for example, a file name or a meta tag) is attached to the image data 20, machine learning is performed without using the text. Therefore, in order to link the two databases (the image information database 200 and the text information database 300), it is necessary to redesign the data format and the data association in each database (the image information database 200 and the text information database 300). Absent. Therefore, while utilizing the current system, it is possible to perform the cooperation between the systems by a simple and flexible method.

  In addition, in the present embodiment, a surgery list in which all or part of the data included in the text data 10 or the column data set 11s and the image data 20 related to the text data 10 or the column data set 11s are associated with each other is generated. It Thereby, for example, a medical staff can reduce misunderstanding of a patient and misunderstanding of a surgical site at the time of surgery.

  Further, in the present embodiment, when the user replaces the image data 20 included in the surgery list with another image data 20, the image data 20 after the replacement and the image corresponding to the image data 20 before the replacement are displayed. Machine learning is performed using the classification name 31 and the learning data. Thereby, the accuracy of learning can be improved.

<2. Modification>
Next, a modification of the information processing system 100 according to the above embodiment will be described.

  In the above embodiment, the number of external databases may be three or more. Further, in the above-described embodiment, the text data 10 is composed of the surgery information, but it may be composed of information other than the surgery information.

  In the above-described embodiment and its modification, the information processing system 100 does not store the data (image data 20) of the image information database 200 in the integrated information database 170 in advance, and the image information database can be used only when necessary. The image data 20 may be read from 200. Similarly, the information processing system 100 does not store the text data 10 (or the column data set 11s) of the text information database 300 in the integrated information database 170 in advance, and from the text information database 300, when necessary, from the text information database 300, The text data 10 (or the column data set 11s) may be read out. In this case, the capacity of the integrated information database 170 can be minimized.

  10, 10A, 10B ... Text data, 11 ... Column data, 11-1, 11-2, 11-3 ... Designated column data, 11c ... Customized column data, 11s ... Column data set, 12 ... Tag, 12-1, 12-2, 12-3 ... Specified tag, 12s ... Tag set, 13 ... Contents, 20, 23 ... Image data, 21 ... Learning image data, 22 ... Test image data, 30 ... Image classification name list, 31 ... Image classification name, 40, 40A, 40B, 40C ... Image classification model, 100 ... Information processing system, 111, 112, 141, 147 ... Screen, 113, 114, 115, 116, 117, 119, 143, 144, 145. 146 ... Button, 118 ... List window, 110, 140 ... GUI, 120 ... Image data IF, 130, 131, 132 ... Image classification learning Program, 150 ... Text data IF, 160 ... Text classification learning program, 171, 172 ... Learning data, 180 ... Information processing unit, 200 ... Image information database, 300 ... Text information database, 1100, 2100, 3100 ... CPU, 1200, 2200, 3200 ... Memory, 1300, 2300, 3300 ... Network IF, 1400 ... Input device, 1500 ... Output device, 2000, 3000 ... Information processing system, 4000 ... Network.

It is a figure showing an example of the functional block of the information processing system concerning one embodiment of the present invention. It is a figure showing an example of the hardware constitutions of the information processing system of FIG. It is a figure showing an example of the text data memorize | stored in the text information database and integrated information database of FIG. It is a figure showing the structural example of the image classification learning program of FIG. It is a figure showing an example of the image data memorize | stored in the image information database and integrated information database of FIG. It is a figure showing an example of the image classification name list memorize | stored in the integrated information database of FIG. It is a figure showing an example of the image data for learning of FIG. It is a figure showing an example of the learning data input into the image classification learning program of FIG. It is a figure showing an example of the learning data input into the image classification learning program of FIG. It is a figure showing an example of the learning data input into the text classification learning program of FIG. It is a figure showing an example of the screen displayed on GUI of FIG. It is a figure showing an example of the screen displayed on GUI of FIG. It is a figure showing an example of the screen displayed on GUI of FIG. It is a figure showing an example of the screen displayed on GUI of FIG. It is a figure showing an example of the data acquisition procedure in the information processing system of FIG. It is a figure showing an example of the machine learning procedure in the information processing system of FIG. It is a figure showing an example of the machine learning procedure in the information processing system of FIG. It is a figure showing an example of the machine learning procedure in the information processing system of FIG. It is a figure showing an example of the test procedure in the information processing system of FIG. It is a figure showing an example of the test procedure in the information processing system of FIG. It is a figure showing an example of the key image acquisition procedure in the information processing system of FIG. It is a figure showing an example of the screen displayed on GUI of FIG.

Claims (9)

A first reading unit that reads the first data from a first database that stores a plurality of first data;
A second reading unit that reads the second data from a second database that stores a plurality of second data;
First machine learning is performed using the first data read from the first database and a first identifier, which is one of a plurality of identifiers, as learning data, and the first data is read from the second database. An information processing system comprising: a learning unit that performs second machine learning using the second data and a second identifier, which is one of a plurality of identifiers, as learning data. The learning unit acquires the first data related to the second data by applying a learning result by the learning unit to the second data read from the second database. 1. The information processing system according to 1. The learning unit generates, as the learning result, information about a correspondence relationship between the second data read from the second database and the identifier by performing the first machine learning, The information processing system according to claim 2, wherein information about a correspondence relationship between the second data read from the second database and the identifier is generated as the learning result by performing machine learning. A storage unit for storing the plurality of first data read from the first database and the plurality of second data read from the second database;
The information processing system according to any one of claims 1 to 3, wherein the learning unit reads the first data related to the second data from the storage unit. The first data includes image data,
The information processing system according to any one of claims 1 to 4, wherein the second data includes a plurality of column data configured by a tag and specific contents of the tag. The information processing system according to claim 5, wherein the plurality of identifiers include a name for classifying the type of the image data included in each of the first data, or a character string or a numerical sequence corresponding to the name. The information processing system according to claim 5, wherein the learning unit performs the first machine learning without using the text even when a text is attached to the image data. The said learning part produces | generates the list | wrist which linked | related all the data contained in the said 2nd data, or one part, and the said 1st data relevant to the said 2nd data. The information processing system described in the item. When the learning unit receives from the user a process of replacing the first data included in the list with another first data, the first data after replacement and the first data before replacement The information processing system according to claim 8, wherein machine learning is performed using the first identifier corresponding to the learning data as learning data.