Information processing device, information processing method, and program

Abstract

An information processing device, wherein a partial graph data generation means generates first partial graph data that corresponds to component data including data pertaining to a plurality of components used for manufacturing one product, and second partial graph data that corresponds to process data including data pertaining to a plurality of processes in the process of manufacturing the one product. An integration means integrates the first partial graph data and the second partial graph data to generate graph data. A training means trains a learning model using the graph data and relationship data that indicates a known relationship between nodes that are linked in the graph data. An analysis means, using the learning model that has been trained, performs an analysis that corresponds to a purpose of analysis related to the manufacturing of the one product. A display information generation means generates display information for showing the analysis results obtained by analysis, together with the grounds thereof.

Description

Technical Field

[0001] The present disclosure relates to a technique for performing processing using graph data.

Background Art

[0002] Graph data such as a knowledge graph has been conventionally known as data capable of expressing relationships between a plurality of entities. Further, graph data has been utilized in recent years for purposes such as improving the productivity of products.

[0003] Specifically, for example, Patent Document 1 discloses a technique of using a graph including nodes representing components of a completed product and edges representing relationships between the components to identify components having the closest relationship to a component selected by a user.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, according to the technique disclosed in Patent Document 1, the relationships between processes in the manufacturing process of a product are not considered. Therefore, according to the technique disclosed in Patent Document 1, there is a problem that, for example, the productivity of the product may decrease due to inconsistencies between processes in the manufacturing process of the product.

[0006] One object of the present disclosure is to provide an information processing apparatus capable of improving the productivity of a product.

Means for Solving the Problems

[0007] In one aspect of this disclosure, the information processing device includes: subgraph data generation means for generating a first subgraph data corresponding to component data including data relating to a plurality of components used in the manufacture of a single product, and a second subgraph data corresponding to process data including data relating to a plurality of processes in the manufacturing process of the single product; integration means for generating graph data by integrating the first subgraph data and the second subgraph data; learning means for training a learning model using the graph data and relationship data which is data indicating known relationships between linked nodes in the graph data; analysis means for performing analysis according to analytical purposes related to the manufacture of the single product using the learned learning model; and display information generation means for generating display information for showing the analysis results obtained from the analysis along with the justification.

[0008] From other perspectives of this disclosure, The computer executes The information processing method generates a first subgraph data corresponding to component data, which includes data relating to multiple components used in the manufacture of a single product, and a second subgraph data corresponding to process data, which includes data relating to multiple processes in the manufacturing process of the single product. It generates graph data by integrating the first subgraph data and the second subgraph data. It trains a learning model using the graph data and relationship data, which is data indicating known relationships between linked nodes in the graph data. It performs analysis according to the analytical purpose related to the manufacture of the single product using the trained learning model, and generates display information to show the analysis results obtained from the analysis along with the justification.

[0009] In other aspects of this disclosure, programThe process involves generating a first partial graph data corresponding to component data, which includes data relating to multiple components used in the manufacture of a single product, and a second partial graph data corresponding to process data, which includes data relating to multiple processes in the manufacturing process of the single product; generating graph data by integrating the first and second partial graph data; training a learning model using the graph data and relationship data, which is data indicating known relationships between linked nodes in the graph data; performing analysis according to the analytical purpose related to the manufacture of the single product using the trained learning model; and generating display information to show the analysis results obtained from the analysis along with the justification. 。 [Effects of the Invention]

[0010] According to this disclosure, it is possible to improve the productivity of the product. [Brief explanation of the drawing]

[0011] [Figure 1] A diagram showing the schematic configuration of an information processing system including a server device according to the first embodiment. [Figure 2] A block diagram showing the hardware configuration of the server device according to the first embodiment. [Figure 3] A block diagram showing the functional configuration of the server device according to the first embodiment. [Figure 4] This diagram shows an example of a record obtained by transforming input data using dictionary data. [Figure 5] Figure 4 shows graph data generated based on each record. [Figure 6] A figure showing an example of graph data obtained by processing by the server device according to the first embodiment. [Figure 7] A diagram illustrating the analysis results obtained by processing by the server device according to the first embodiment. [Figure 8] A diagram showing example data obtained by processing by the server device according to the first embodiment. [Figure 9] A diagram showing display information generated by the processing of the server device according to the first embodiment. [Figure 10] A flowchart showing an example of the processing performed in the server device according to the first embodiment. [Figure 11] A block diagram showing the functional configuration of the information processing device according to the second embodiment. [Figure 12] A flowchart for explaining the processing performed in the information processing device according to the second embodiment.

Mode for Carrying Out the Invention

[0012] Hereinafter, preferred embodiments of the present disclosure will be described with reference to the drawings.

[0013] <First Embodiment> [System Configuration] FIG. 1 is a diagram showing a schematic configuration of an information processing system including a server device according to the first embodiment. As shown in FIG. 1, the information processing system 1 includes a server device 100 and a terminal device 200.

[0014] The server device 100 is configured to be able to communicate with the terminal device 200. Further, the server device 100 performs analysis using a learning model described later, and thereby obtains, for example, analysis results related to defects that may occur due to design changes and / or process changes of products. Further, the server device 100 generates display information for showing the above-mentioned analysis results together with the basis, and outputs the generated display information to the terminal device 200.

[0015] The terminal device 200 has functions such as a function of communicating with the server device 100, a function of inputting information transmitted to the server device 100, and a function of displaying information received from the server device 100. Further, the terminal device 200 has a function of displaying information according to a user's operation. Specifically, the terminal device 200 may be configured by a device such as a personal computer, a smartphone, and a tablet computer.

[0016] [Hardware configuration] Figure 2 is a block diagram showing the hardware configuration of a server device according to the first embodiment. As shown in Figure 2, the server device 100 includes an interface (IF) 111, a processor 112, a memory 113, a recording medium 114, and a database (DB) 115.

[0017] IF111 performs data input and output with external devices. For example, information transmitted from terminal device 200 is input to server device 100 through IF111.

[0018] The processor 112 is a computer such as a CPU (Central Processing Unit) and controls the entire server device 100 by executing pre-prepared programs. Specifically, the processor 112 performs analysis using the learning model described later, for example.

[0019] Memory 113 consists of ROM (Read Only Memory), RAM (Random Access Memory), and other components. Memory 113 is also used as working memory while the processor 112 is executing various processes.

[0020] The recording medium 114 is a non-volatile, non-temporary recording medium such as a disk-shaped recording medium or semiconductor memory, and is configured to be detachable from the server device 100. The recording medium 114 stores various programs that the processor 112 executes. When the server device 100 performs various processes, the programs stored in the recording medium 114 are loaded into the memory 113 and executed by the processor 112.

[0021] DB115 stores, for example, information input through IF111 and processing results obtained from the processor 112.

[0022] [Functional Configuration] Figure 3 is a block diagram showing the functional configuration of a server device according to the first embodiment. As shown in Figure 3, the server device 100 includes an analysis data storage unit 11, a graph data generation unit 12, a graph data storage unit 13, a learning processing unit 14, a processing result storage unit 15, a query generation unit 16, an analysis processing unit 17, and a display information generation unit 18.

[0023] The analysis data storage unit 11 stores component data ADP, process data ADK, and dictionary data JD.

[0024] The component data ADP contains data relating to multiple components used in the manufacture of a single product. Specifically, the component data ADP includes data indicating the attributes of each of the multiple components used in the manufacture of product Y, which corresponds to the finished product, and the relationships between those components. Component attributes may include, for example, the shape, material, and size of the component. Relationships between components may include, for example, the hierarchical relationships between the multiple components with product Y as the top level. Note that the aforementioned hierarchical relationships may also be interpreted as parent-child relationships.

[0025] The process data ADK contains data related to multiple processes in the manufacturing process of a single product. Specifically, the process data ADK includes data indicating the attributes of each of the multiple processes in the manufacturing process of product Y, which is the finished product, and the relationships between those processes. Process attributes may include, for example, the content of the work and the entity (person or equipment) that performs the work. Relationships between processes may include, for example, the sequence of multiple processes in the manufacturing process of product Y.

[0026] The dictionary data JD contains data indicating the rules that are referenced in the generation of graph data by the graph data generation unit 12 and the generation of queries by the query generation unit 16.

[0027] The graph data generation unit 12 generates graph data GD based on the component data ADP, process data ADK, and dictionary data JD read from the analysis data storage unit 11, and stores the generated graph data GD in the graph data storage unit 13. The graph data generation unit 12 also includes a data conversion unit 12A and an integrated processing unit 12B.

[0028] The data conversion unit 12A functions as a means for generating subgraph data. The data conversion unit 12A converts component data ADP into subgraph data GDP based on rules indicated by dictionary data JD. The data conversion unit 12A also converts process data ADK into subgraph data GDK based on rules indicated by dictionary data JD.

[0029] The integrated processing unit 12B functions as an integration means. The integrated processing unit 12B also generates graph data GD by integrating the partial graph data GDP and GDK, and stores the generated graph data GD in the graph data storage unit 13.

[0030] The graph data storage unit 13 stores the graph data GD generated by the graph data generation unit 12.

[0031] The learning processing unit 14 functions as a learning means. The learning processing unit 14 also acquires relationship data KD, which represents known relationships between linked nodes in the graph data GD read from the graph data storage unit 13. The relationship data KD contains data that shows the relationships between multiple nodes as a universal rule by replacing each of the multiple nodes that have the same relationship with each other with a variable node. The learning processing unit 14 then uses the graph data GD and the relationship data KD to train the learning model GMD so that unknown relationships between unlinked nodes in the graph data GD are derived. The learning processing unit 14 then stores the relationship data KD and the trained learning model GMD in the processing result storage unit 15.

[0032] The processing result storage unit 15 stores the relationship data KD and the trained learning model GMD as data obtained by the learning processing unit 14.

[0033] The query generation unit 16 functions as a query generation means. Furthermore, the query generation unit 16 generates queries corresponding to the analysis purpose input from the terminal device 200, based on rules indicated by the dictionary data JD. The analysis purpose can be set, for example, to something related to the manufacturing of product Y.

[0034] The analysis processing unit 17 functions as an analysis tool. The analysis processing unit 17 uses the trained learning model GMD read from the processing result storage unit 15 to perform analysis on queries generated by the query generation unit 16. The analysis processing unit 17 also outputs the analysis results obtained from the aforementioned analysis to the display information generation unit 18.

[0035] The display information generation unit 18 functions as a means for generating display information. The display information generation unit 18 extracts rules from among the rules contained in the relationship data KD read from the processing result storage unit 15 that form the basis of the analysis results obtained by the analysis processing unit 17. The display information generation unit 18 also identifies the portion of the graph data GD read from the graph data storage unit 13 in which the extracted rules are valid, and obtains example data ED corresponding to the data of the identified portion. The display information generation unit 18 then generates display information HJ based on the analysis results obtained by the analysis processing unit 17 and the example data ED, and outputs the generated display information HJ to the terminal device 200.

[0036] [Specific example] Next, we will explain specific examples of the processes performed in each part of the server device 100.

[0037] The dictionary data JD contains rules for converting input data in various formats, such as tabular data, into records of a predetermined format. According to these rules, for example, input data can be converted into records RA~RD as shown in Figure 4. Figure 4 shows an example of a record obtained by converting input data using the dictionary data.

[0038] For each of records RA through RD, the leftmost element (the strings "Jacket", "T-Shirt", and "Tom") represents the source node in the graph, and the rightmost element (the strings "clothes" and "Male") represents the destination node in that graph. Furthermore, for each of records RA through RD, the central element (the strings "is_category", "is_gender", and "interested_for") represents the relationship between the linked nodes in the graph. Additionally, for each of records RA through RD, the order of each element (each string) represents the link direction (the direction of the arrows in the graph) when linking nodes in the graph. Therefore, records RA through RD in Figure 4 can be treated as the basis data for the graph data GDX shown in Figure 5. Note that the arrows linking nodes in the graph can be interpreted as edges. Figure 5 shows the graph data generated based on each record in Figure 4.

[0039] The data conversion unit 12A generates subgraph data GDP containing multiple records corresponding to the part data ADP by converting the part data ADP based on the rules described in the above example while referring to the dictionary data JD. The data conversion unit 12A also generates subgraph data GDK containing multiple records corresponding to the process data ADK by converting the process data ADK based on the rules described in the above example while referring to the dictionary data JD. In other words, the data conversion unit 12A generates a first subgraph data corresponding to the part data and a second subgraph data corresponding to the process data. Furthermore, the data conversion unit 12A generates the first and second subgraph data by converting each of the part data and process data into data in a predetermined format that represents four elements: the source node, the destination node, the relationship between the nodes, and the link direction when linking between nodes.

[0040] The integrated processing unit 12B generates graph data GD by integrating partial graph data GDP and GDK, and stores the generated graph data GD in the graph data storage unit 13. Specifically, the integrated processing unit 12B generates graph data GD as shown in Figure 6 by integrating the partial graph data GDP and GDK based on common elements contained in the partial graph data GDP and GDK, and stores the generated graph data GD in the graph data storage unit 13. Figure 6 is a diagram showing an example of graph data obtained by processing of the server device according to the first embodiment.

[0041] The graph data GD in Figure 6 is generated, for example, when the data obtained when part P used in the manufacture of product Y is updated to part Q is included in the part data ADP that forms the basis of the partial graph data GDP, and the data obtained when process K in the manufacturing process of product Y is updated to process L is included in the part data ADK that forms the basis of the partial graph data GDK. The graph data GD in Figure 6 includes a "Part P" node, a "Part Q" node, a "Process K" node, and a "Part L" node. The graph data GD in Figure 6 also includes a "Part Type" node indicating that part P and part Q are the same type of part. The graph data GD in Figure 6 also includes a "Process Attribute" node indicating that process K and process L have the same attributes. The graph data GD in Figure 6 also includes a "Design Change ID" node representing the design change from part P to part Q, and a "Process Change ID" node representing the process change from process K to process L. Furthermore, the graph data GD in Figure 6 includes a "Defect DX" node representing common defects that occur when part R, which was used in the manufacturing of product Y, is changed to part Q, and when process M, which was included in the manufacturing process of product Y, is changed to process K. For simplicity, in this specific example, the diagrams of nodes corresponding to part R and process M mentioned above have been omitted. The graph data GD in Figure 6 also includes a "Higher-Level Part" node representing a higher-level part relative to part P, and a "Lower-Level Part" node representing a lower-level part relative to part P. The graph data GD in Figure 6 also includes a "Previous Process" node representing the process immediately preceding process L, and a "Next Process" node representing the process immediately following process L. The arrow from the "Part Q" node to the "Process K" node in the graph data GD in Figure 6 indicates that process K is performed using part Q.

[0042] The learning processing unit 14 obtains relationship data KD, which indicates known relationships between linked nodes in the graph data GD read from the graph data storage unit 13.

[0043] The relationship data KD includes data that shows the relationships corresponding to each of the multiple arrows in the graph data GD in Figure 6. Specifically, the relationship data KD includes, for example, data that shows the rule that defect DX occurs when one of the parts used in the manufacture of product Y is changed to part Q, and data that shows the rule that defect DX occurs when the manufacturing process of product Y includes a process that has the same attributes as process K.

[0044] The learning processing unit 14 trains the learning model GMD, which is constructed based on "KBLRN," by inputting the feature quantities corresponding to each node in the graph data GD and the feature quantities corresponding to each relationship in the relationship data KD. The learning processing unit 14 also stores the relationship data KD and the trained learning model GMD in the processing result storage unit 15.

[0045] The aforementioned "KBLRN" is disclosed, for example, in Alberto Garcia-Duran, et al., "KBLRN: End-to-End Learning of Knowledge Base Representations with Latent, Relational, and Numerical Features". Furthermore, the learning model GMD may be constructed based on a model other than "KBLRN", as long as it has a configuration that enables link prediction for graph data. In addition, when the learning processing unit 14 trains the learning model GMD using graph data GD and relational data KD, it is desirable to perform zero-shot training, such as that disclosed in Japanese Patent Application Publication No. 2019-125364. Furthermore, according to this specific example, for example, each time at least one of the component data ADP and process data ADK is updated, the graph data GD may be updated, and the learning model GMD may be retrained using the updated graph data GD.

[0046] The query generation unit 16 generates a record based on the rules indicated by the dictionary data JD, corresponding to the analysis objective input from the terminal device 200. For example, the record represents the source node, the relationship between nodes, and the link direction, while the destination node is unknown. For example, when the analysis objective input is "predicting potential defects that may occur when a design change is made from part P to part Q in the manufacturing of product Y," the query generation unit 16 generates a query QA corresponding to that analysis objective. For example, query QA is generated as a record that contains a string representing "design change from part P to part Q" corresponding to the source node, a string representing "potential defects" corresponding to the relationship between nodes, and the word order of "design change from part P to part Q" and "potential defects" corresponding to the link direction is clear, but there is no character or string representing the destination node. The query generation unit 16 also generates a query QB corresponding to the analysis objective input is "predicting potential defects that may occur when a process change is made from process K to process L in the manufacturing process of product Y." Query QB is generated as a record in which, for example, there is a string representing "process change from process K to process L" corresponding to the source node, and a string representing "possible defects" corresponding to the relationship between nodes, and the word order of "process change from process K to process L" and "possible defects" corresponding to the link direction is clear, but there is no character or string representing the destination node. The query generation unit 16 only needs to generate a record in which one of the four elements—the source node, the destination node, the relationship between nodes, and the link direction—is unknown, as a query according to the analysis purpose input from the terminal device 200.

[0047] The analysis processing unit 17 uses the trained learning model GMD read from the processing result storage unit 15 to perform analysis on the query generated by the query generation unit 16. Furthermore, by using the trained learning model GMD to perform analysis on the query generated by the query generation unit 16, the analysis processing unit 17 obtains an analysis result corresponding to one unknown element in the query. Finally, the analysis processing unit 17 outputs the analysis results obtained from the aforementioned analysis to the display information generation unit 18.

[0048] For example, the analysis processing unit 17 uses a trained learning model GMD to perform an analysis to predict the linked node in the query QA record. Through such an analysis, the analysis processing unit 17 can obtain an analysis result corresponding to the dashed arrow pointing from the "design change ID" node to the "defect DX" node in the graph data GD, as shown in Figure 7. In other words, by using the trained learning model GMD to perform an analysis related to the query QA, the analysis processing unit 17 can obtain an analysis result indicating that defect DX may occur if part P used in the manufacture of product Y is changed to part Q. Figure 7 is a diagram illustrating the analysis results obtained by the processing of the server device according to the first embodiment.

[0049] Furthermore, for example, the analysis processing unit 17 uses the trained learning model GMD to perform analysis to predict the linked node in the record of query QB. Through such analysis, the analysis processing unit 17 can obtain an analysis result corresponding to the dashed arrow pointing from the "Process Change ID" node to the "Defect DX" node in the graph data GD, as shown in Figure 7. In other words, by using the trained learning model GMD to perform analysis related to query QB, the analysis processing unit 17 can obtain an analysis result indicating that defect DX may occur if process K, which is included in the manufacturing process of product Y, is changed to process L.

[0050] The display information generation unit 18 acquires example data EDA as shown in Figure 8 when, for example, graph data GD as shown in Figure 6 is stored in the graph data storage unit 13 and analysis results corresponding to query QA are obtained by the analysis processing unit 17. Figure 8 is a diagram showing example data obtained by processing of the server device according to the first embodiment.

[0051] The example data EDA in Figure 8 corresponds to the rules shown by the relationship data KD read from the processing result storage unit 15 in the graph data GD in Figure 6, and is the data that has a high correlation with the analysis results corresponding to query QA.

[0052] The display information generation unit 18 generates display information HJA, for example, as shown in Figure 9, based on the example data EDA and the analysis results corresponding to the query QA, and outputs the generated display information HJA to the terminal device 200. Figure 9 is a diagram showing the display information generated by the processing of the server device according to the first embodiment.

[0053] The HJA display information in Figure 9 is composed of a graph containing the following nodes: "Part Q," "Design Change P⇒Q," "Design Change R⇒Q," and "Defect DX." Each node in the HJA display information in Figure 9 is linked by arrows indicating the relationships between nodes. The "Design Change P⇒Q" node indicates the analysis objective that formed the basis of the query QA, and shows the design change from part P to part Q in product Y. The dashed arrow from the "Design Change P⇒Q" node to the "Defect DX" node shows the analysis result corresponding to the query QA, and indicates that a defect DX may occur due to the design change from part P to part Q. The "Design Change R⇒Q" node indicates a design change from part R to part Q in product Y that occurred prior to the design change from part P to part Q in product Y. The solid arrow from the "Design Change R⇒Q" node to the "Defect DX" node indicates that a defect DX occurred in the past due to the design change from part R to part Q. Therefore, according to the display information HJA in Figure 9, it is possible to show, based on a past case in which defect DX occurred when part R in product Y was changed to part Q, that defect DX could also occur if part P in product Y were changed to part Q. In other words, the display information generation unit 18 can generate display information HJA to show the analysis results along with the basis, based on the example data EDA and the analysis results corresponding to the query QA. Furthermore, the display information generation unit 18 can generate display information to show, as an analysis result, that defect DX could occur due to the first design change (change from part P to part Q), based on the fact that defect DX occurred due to a second design change (change from part R to part Q) that was carried out before the first design change (change from part P to part Q) corresponding to the analysis objective.

[0054] Furthermore, according to this specific example, the display information generation unit 18 may generate display information that includes comments such as, "There is a possibility that a malfunction DX may occur if part P is changed to part Q."

[0055] Furthermore, the display information generation unit 18 can generate display information HJB, which shows the analysis results corresponding to query QB along with the justification, by performing the same processing as the processing related to the generation of the display information HJA described above. According to the display information HJB, it is possible to show, as justification, a past case in which defect DX occurred when process M in the manufacturing process of product Y was changed to process K, and to show a prediction that defect DX may also occur when process K in the manufacturing process of product Y is changed to process L. In other words, the display information generation unit 18 can show, as justification, that defect DX occurred due to a second process change (change from process M to process K) that was performed before the first process change (change from process K to process L) corresponding to the analysis objective, and to show as an analysis result that defect DX may occur due to the first process change.

[0056] [Processing flow] Next, the processing flow performed in the server device according to the first embodiment will be described. Figure 10 is a flowchart showing an example of the processing performed in the server device according to the first embodiment.

[0057] First, the server device 100 generates subgraph data corresponding to the component data and process data stored in the analysis data storage unit 11 (step S11).

[0058] Next, the server device 100 generates graph data by integrating the subgraph data generated in step S11 (step S12).

[0059] Next, the server device 100 trains the learning model using the graph data generated in step S12 and relationship data indicating known relationships between linked nodes in the graph data (step S13).

[0060] Next, the server device 100 generates a query corresponding to the analysis purpose input from the terminal device 200 (step S14).

[0061] Next, the server device 100 uses the trained model obtained in step S13 to perform an analysis on the query generated in step S14 (step S15).

[0062] Next, the server device 100 generates display information to show the analysis results obtained in step S15 along with the supporting evidence (step S16).

[0063] Next, the server device 100 outputs the display information generated in step S16 to the terminal device 200 (step S17).

[0064] As described above, according to this embodiment, a first partial graph data corresponding to part data and a second partial graph data corresponding to process data are generated, and the learning model is trained using graph data that integrates the first partial graph data and the second graph data. Furthermore, according to this embodiment, analysis can be performed according to the analytical purpose related to product manufacturing using the learning model that has been trained as described above. Furthermore, according to this embodiment, the analysis results obtained from the above analysis can be presented to the user along with the rationale. Specifically, according to this embodiment, for example, analysis results that predict defects that may occur due to the relationships between parts and / or between processes can be presented to the user along with the rationale. Therefore, according to this embodiment, the occurrence of defects shown in the above analysis results can be prevented, thereby improving product productivity. Furthermore, according to this embodiment, the rationale for the defects shown in the above analysis results can be grasped, so appropriate countermeasures to prevent the occurrence of such defects can be formulated. Furthermore, according to this embodiment, by performing analysis using the above-described learning model, highly accurate analysis results according to the analytical purpose can be obtained.

[0065] <Second Embodiment> Figure 11 is a block diagram showing the functional configuration of the information processing device according to the second embodiment.

[0066] The information processing device 500 according to this embodiment has the same hardware configuration as the server device 100. The information processing device 500 also includes a partial graph data generation means 511, an integration means 512, a learning means 513, an analysis means 514, and a display information generation means 515.

[0067] Figure 12 is a flowchart illustrating the processing performed in the information processing device according to the second embodiment.

[0068] The subgraph data generation means 511 generates a first subgraph data corresponding to component data, which includes data relating to multiple components used in the manufacture of a single product, and a second subgraph data corresponding to process data, which includes data relating to multiple processes in the manufacturing process of the single product (step S51).

[0069] The integration means 512 generates graph data by integrating the first partial graph data and the second partial graph data (step S52).

[0070] The learning means 513 trains the learning model 600 using graph data and relationship data, which is data indicating known relationships between linked nodes in the graph data (step S53).

[0071] The analysis means 514 uses the trained model 600 to perform an analysis according to the analytical purpose related to the manufacture of a product (step S54).

[0072] The display information generation means 515 generates display information to show the analysis results obtained through the analysis, along with the supporting evidence (step S55).

[0073] According to this embodiment, the productivity of the product can be improved.

[0074] Some or all of the above embodiments may also be described as follows, but are not limited to the following:

[0075] (Note 1) A subgraph data generation means that generates a first subgraph data corresponding to component data including data relating to multiple components used in the manufacture of a single product, and a second subgraph data corresponding to process data including data relating to multiple processes in the manufacturing process of the single product, An integration means for generating graph data by integrating the first partial graph data and the second partial graph data, A learning means for training a learning model using the aforementioned graph data and relationship data which is data indicating known relationships between linked nodes in the aforementioned graph data, An analytical means that uses the learned model on which the learning has been performed to perform an analysis according to the analytical purpose related to the manufacture of the one product, A display information generation means for generating display information to show the analysis results obtained from the above analysis along with the supporting evidence, An information processing device having

[0076] (Note 2) The subgraph data generation means is an information processing device that generates the first subgraph data and the second subgraph data by converting each of the component data and the process data into data in a predetermined format that represents four elements: the source node, the destination node, the relationship between the nodes, and the linking direction when linking the nodes. (Appendix 1)

[0077] (Note 3) The system further includes a query generation means that generates data in which one of the four elements is unknown, as a query corresponding to the aforementioned analysis objective. The analysis means is an information processing device described in Appendix 2 that obtains an analysis result corresponding to one of the elements by performing an analysis related to the query using the learning model on which the learning has been performed.

[0078] (Note 4) The integration means is an information processing device described in Appendix 1 that generates graph data by integrating the plurality of subgraph data based on common elements contained in the plurality of subgraph data.

[0079] (Note 5) The learning means is an information processing device described in Appendix 1 that trains the learning model so that unknown relationships between unlinked nodes in the graph data can be derived.

[0080] (Note 6) The analysis means uses the learned model on which the learning has been performed to perform an analysis to predict defects that may occur when a first design change is made to the one product. The display information generation means is an information processing device for Appendix 1 that generates display information to indicate, as an analysis result, that a defect may occur due to the first design change, while indicating, as evidence, that a defect occurred due to a second design change that was carried out before the first design change.

[0081] (Note 7) The analysis means uses the learned model on which the learning has been performed to perform an analysis to predict defects that may occur when a first process change is made in one of the products. The display information generation means is an information processing device for Appendix 1 that generates display information to indicate, as an analysis result, that a defect may occur due to the first process change, while indicating, as evidence, that a defect occurred due to a second process change that was performed before the first process change.

[0082] (Note 8) The system generates a first partial graph data corresponding to component data, which includes data related to multiple components used in the manufacture of a single product, and a second partial graph data corresponding to process data, which includes data related to multiple processes in the manufacturing process of the single product. Graph data is generated by integrating the first partial graph data and the second partial graph data. The learning model is trained using the aforementioned graph data and relationship data, which is data indicating known relationships between linked nodes in the aforementioned graph data. Using the learned model on which the aforementioned learning has been performed, an analysis is conducted according to the analytical purpose related to the manufacturing of the one product. An information processing method for generating display information to show the analysis results obtained from the above analysis, along with the supporting evidence.

[0083] (Note 9) The system generates a first partial graph data corresponding to component data, which includes data related to multiple components used in the manufacture of a single product, and a second partial graph data corresponding to process data, which includes data related to multiple processes in the manufacturing process of the single product. Graph data is generated by integrating the first partial graph data and the second partial graph data. The learning model is trained using the aforementioned graph data and relationship data, which is data indicating known relationships between linked nodes in the aforementioned graph data. Using the learned model on which the aforementioned learning has been performed, an analysis is conducted according to the analytical purpose related to the manufacturing of the one product. A recording medium containing a program that causes a computer to execute a process to generate display information for showing the analysis results obtained from the aforementioned analysis, along with the supporting evidence.

[0084] Although the present disclosure has been described above with reference to embodiments, the present disclosure is not limited to the above embodiments and examples. Various modifications to the structure and details of the present disclosure are possible, as can be understood by those skilled in the art within the scope of the present disclosure. [Explanation of Symbols]

[0085] 12 Graph Data Generation Unit 12A Data Conversion Unit 12B Integrated Processing Unit 14. Learning Processing Unit 16. Query Generation Section 17. Analytical Processing Section 18 Display information generation section 100 Server Devices

Claims

1. A subgraph data generation means that generates a first subgraph data corresponding to component data including data relating to multiple components used in the manufacture of a single product, and a second subgraph data corresponding to process data including data relating to multiple processes in the manufacturing process of the single product, An integration means for generating graph data by integrating the first partial graph data and the second partial graph data, A learning means for training a learning model using the aforementioned graph data and relationship data which is data indicating known relationships between linked nodes in the aforementioned graph data, An analytical means that uses the learned model on which the learning has been performed to perform an analysis according to the analytical purpose related to the manufacture of the one product, A display information generation means for generating display information to show the analysis results obtained from the above analysis along with the supporting evidence, An information processing device having

2. The information processing apparatus according to claim 1, wherein the subgraph data generation means generates the first subgraph data and the second subgraph data by converting each of the component data and the process data into data in a predetermined format that represents four elements: the source node, the destination node, the relationship between the nodes, and the linking direction when linking the nodes.

3. The system further includes a query generation means that generates data in which one of the four elements is unknown, as a query corresponding to the aforementioned analysis objective. The information processing apparatus according to claim 2, wherein the analysis means obtains an analysis result corresponding to one element by performing an analysis related to the query using the learning model on which the learning has been performed.

4. The information processing apparatus according to claim 1, wherein the integration means generates graph data by integrating the plurality of subgraph data based on common elements contained in the plurality of subgraph data.

5. The information processing apparatus according to claim 1, wherein the learning means trains the learning model such that unknown relationships between unlinked nodes in the graph data are derived.

6. The analysis means uses the learned model on which the learning has been performed to perform an analysis to predict defects that may occur when a first design change is made to the one product. The information processing apparatus according to claim 1, wherein the display information generating means generates display information to indicate, as the basis, that a defect occurred due to a second design change made prior to the first design change, and as the analysis result, that the defect may occur due to the first design change.

7. The analysis means uses the learned model on which the learning has been performed to perform an analysis to predict defects that may occur when a first process change is made in one of the products. The information processing apparatus according to claim 1, wherein the display information generating means generates display information to indicate, as the basis, that a defect occurred due to a second process change performed before the first process change, and as the analysis result, that the defect may occur due to the first process change.

8. A computer-based information processing method, A first partial graph data is generated that corresponds to component data, which includes data related to multiple components used in the manufacture of a single product, and a second partial graph data is generated that corresponds to process data, which includes data related to multiple processes in the manufacturing process of the single product. Graph data is generated by integrating the first partial graph data and the second partial graph data. The learning model is trained using the aforementioned graph data and relationship data, which is data indicating known relationships between linked nodes in the aforementioned graph data. Using the learned model on which the aforementioned learning has been performed, an analysis is conducted according to the analytical purpose related to the manufacturing of the one product. An information processing method for generating display information to show the analysis results obtained from the above analysis, along with the supporting evidence.

9. A first partial graph data is generated that corresponds to component data, which includes data related to multiple components used in the manufacture of a single product, and a second partial graph data is generated that corresponds to process data, which includes data related to multiple processes in the manufacturing process of the single product. Graph data is generated by integrating the first partial graph data and the second partial graph data. The learning model is trained using the aforementioned graph data and relationship data, which is data indicating known relationships between linked nodes in the aforementioned graph data. Using the learned model on which the aforementioned learning has been performed, an analysis is conducted according to the analytical purpose related to the manufacturing of the one product. A program that causes a computer to execute a process to generate display information that shows the analysis results obtained from the aforementioned analysis, along with the supporting evidence.

Images