Data model generation device, data model generation system, and data model generation method

Abstract

The purpose of the present invention is to provide a technique capable of generating an appropriate data model even without domain information such as detailed classification conditions. A data model generation device is provided with: an attribute candidate generation unit that generates attribute candidates from a plurality of pieces of selection attribute information on the basis of a plurality of pieces of evaluation values for the selection attribute information obtained by repeating selection for the selection attribute information and calculation of the evaluation values for the selection attribute information; and a data model editing unit that updates the data model on the basis of the tag information and the attribute candidates or on the basis of the attribute candidates.

Description

Technical Field

[0001] This disclosure relates to a data model generation apparatus, a data model generation system, and a data model generation method. Background Technology

[0002] To improve information retrieval or visualize the relationships between data, a data model using graph structures to represent these relationships (hereafter sometimes abbreviated as "model") has been proposed. This data model is used in areas such as social graphs, recommendation systems, geospatial systems, and master data management.

[0003] In large-scale facilities such as water treatment plants and power plants, the number of plant components under monitoring is enormous, resulting in a large size of data models used to manage the data and alarms collected from these components. Manually creating, managing, and maintaining such data models requires a significant amount of labor. Furthermore, as the model grows larger, the relationships between data important to users referencing the model are obscured, impairing the model's visibility.

[0004] Regarding labor in such problems, methods are proposed to generate models based on existing domain information such as engineering diagrams and knowledge graphs, as well as methods to extend existing small-scale models. Regarding model visibility, the following methods are proposed: improving the visibility of large-scale models by dividing the structural elements of the model into predetermined categories or by linking and aggregating related elements.

[0005] For example, Patent Document 1 discloses a technique that generates a data model by associating factory components extracted from engineering diagrams and other engineering sources based on predetermined classification conditions.

[0006] Existing technical documents

[0007] Patent Document 1: Japanese Patent Publication No. 2022-524642 Summary of the Invention

[0008] However, in existing technologies, detailed design of classification criteria, i.e., domain information, is required. Therefore, when hierarchical structuring is performed based on multiple categories, the design time for domain information increases. On the other hand, without detailed design of domain information, it becomes difficult to categorize based on user preferences.

[0009] Therefore, this disclosure was made in view of the aforementioned problems, and its purpose is to provide a technique that can generate appropriate data models even without detailed information in areas such as classification conditions.

[0010] The data model generation apparatus disclosed herein comprises: an attribute information extraction and management unit that extracts a set of attribute information from a data model of a monitored object; a selection unit that selects a portion of the attribute information as selected attribute information from the set of attribute information; an evaluation unit that calculates an evaluation value for the selected attribute information; an attribute candidate generation unit that generates attribute candidates based on the evaluation values ​​of a plurality of selected attribute information obtained by repeatedly selecting the selected attribute information and calculating the evaluation values ​​of the selected attribute information; and a data model editing unit that updates the data model based on label information associated with a classification label representing the attribute candidate and the attribute candidate, or based on the attribute candidate.

[0011] According to this disclosure, based on the evaluation values ​​of multiple selected attribute information obtained by repeatedly selecting and calculating the evaluation values ​​of the selected attribute information, attribute candidates are generated according to the multiple selected attribute information, and the data model is updated according to the label information and the attribute candidates, or according to the attribute candidates. With this structure, an appropriate data model can be generated even without detailed information such as classification conditions.

[0012] The purpose, features, solutions, and advantages of this disclosure will become clearer from the following detailed description and accompanying drawings. Attached Figure Description

[0013] Figure 1 This is a diagram showing the structure of the data model generation apparatus according to Embodiment 1.

[0014] Figure 2 This is a flowchart illustrating a summary of the processing procedure of the data model generation apparatus according to Embodiment 1.

[0015] Figure 3 This is a diagram illustrating an example of attribute information extraction information and attribute information selection evaluation information involved in Implementation 1.

[0016] Figure 4 This is a diagram illustrating an example of the data stored in the data model information storage unit according to Embodiment 1.

[0017] Figure 5 This is a flowchart illustrating the processing procedure of the attribute information extraction management unit involved in Implementation Method 1.

[0018] Figure 6This is a diagram illustrating the operation of the attribute information extraction management unit involved in Implementation Method 1.

[0019] Figure 7 This is a diagram illustrating the operation of the attribute information extraction management unit involved in Implementation Method 1.

[0020] Figure 8 This is a diagram illustrating the operation of the attribute information extraction management unit involved in Implementation Method 1.

[0021] Figure 9 This is a diagram illustrating an example of the information managed by the attribute information extraction management department according to Embodiment 1.

[0022] Figure 10 This is a flowchart illustrating the processing procedure of the attribute candidate generation unit according to Embodiment 1.

[0023] Figure 11 This is a diagram used to illustrate the operation of the attribute candidate generation unit involved in Embodiment 1.

[0024] Figure 12 This is a diagram used to illustrate the operation of the attribute candidate generation unit involved in Embodiment 1.

[0025] Figure 13 This is a flowchart illustrating the processing procedure of the selection unit involved in Embodiment 1.

[0026] Figure 14 This is a diagram used to explain the operation of the selection unit involved in Embodiment 1.

[0027] Figure 15 This is a flowchart illustrating the processing procedure of the evaluation unit involved in Embodiment 1.

[0028] Figure 16 This is a diagram used to illustrate the operation of the evaluation unit involved in Embodiment 1.

[0029] Figure 17 This is a flowchart illustrating the processing procedure of the data model editing unit involved in Implementation Method 1.

[0030] Figure 18 This is a diagram illustrating the operation of the data model editing unit involved in Implementation Method 1.

[0031] Figure 19 This is a diagram illustrating the operation of the data model editing unit involved in Implementation Method 1.

[0032] Figure 20 This is a diagram showing the structure of the data model generation apparatus according to Embodiment 2.

[0033] Figure 21This is a flowchart illustrating the processing procedure of the attribute candidate generation unit according to Embodiment 2.

[0034] Figure 22 This is a diagram used to illustrate the operation of the attribute candidate generation unit involved in Embodiment 2.

[0035] Figure 23 This is a flowchart illustrating the processing procedure of the selection unit involved in Embodiment 2.

[0036] Figure 24 This is a diagram showing the structure of the data model generation apparatus according to Embodiment 3.

[0037] Figure 25 This is a flowchart illustrating the processing procedure of the attribute candidate generation unit according to Embodiment 3.

[0038] Figure 26 This is a flowchart illustrating the processing procedure of the evaluation learning unit involved in Implementation Method 3.

[0039] Figure 27 This is a diagram showing the structure of the data model generation apparatus according to Embodiment 4.

[0040] Figure 28 This is a diagram used to illustrate the operation of the attribute candidate generation unit involved in Embodiment 4.

[0041] Figure 29 This is a flowchart illustrating the processing procedure of the evaluation learning unit involved in Implementation 4.

[0042] Figure 30 This is a flowchart illustrating the processing procedure of the selection learning unit involved in Embodiment 4.

[0043] Figure 31 This is a diagram showing the structure of the data model generation apparatus according to Embodiment 5.

[0044] Figure 32 This is a flowchart illustrating the processing procedure of the selection learning unit involved in Implementation 5.

[0045] Figure 33 This is a block diagram illustrating the hardware structure of the data model generation apparatus involved in other variations.

[0046] Figure 34 This is a block diagram illustrating the hardware structure of the data model generation apparatus involved in other variations. Detailed Implementation

[0047] <Implementation Method 1>

[0048] Figure 1This is a diagram showing the structure of the data model generation apparatus according to Embodiment 1. Figure 1 The data model generation device includes a data model information storage unit 101, an attribute information extraction and management unit 102, an extraction setting storage unit 103, an attribute candidate generation unit 104, a setting acceptance unit 107, and a data model editing unit 108. The attribute candidate generation unit 104 includes a selection unit 105 and an evaluation unit 106.

[0049] As explained below, the data model generation device extracts a set of attribute information from the data model of the monitored object, aggregates the set of attribute information to generate attribute candidates, and associates the category labels representing the attribute candidates with the attribute candidates, thereby updating the data model. Thus, even without detailed information such as classification criteria, a new data model with visibility and searchability can be generated.

[0050] Figure 2 This is a flowchart illustrating a summary of the processing procedure of the data model generation apparatus according to Embodiment 1.

[0051] In step S201, the receiving unit 107 receives information from external users, including attribute information extraction information and attribute information selection evaluation information, and records it in the extraction setting storage unit 103. Details will be described later. Attribute information extraction information is information used to obtain a data model or extract a set of attribute information from a data model, and attribute information selection evaluation information is information used to select attribute information or evaluate the selected attribute information.

[0052] In step S202, the attribute information extraction management unit 102 obtains attribute information extraction information from the extraction setting storage unit 103. The attribute information extraction management unit 102 obtains the corresponding data model from the data model information storage unit 101 according to the data model name specified in the attribute information extraction information.

[0053] In step S203, the attribute information extraction management unit 102 extracts a set of attribute information from the data model obtained in step S202 based on the attribute information extraction information obtained in step S202 and manages it.

[0054] In step S204, the attribute candidate generation unit 104 obtains the set of attribute information managed by the attribute information extraction management unit 102, and obtains attribute information selection evaluation information from the extraction setting storage unit 103.

[0055] In step S205, the attribute candidate generation unit 104 selects a partial set of attribute information from the set of attribute information to become the selected attribute information based on the attribute selection evaluation information obtained in step S204. That is, the attribute candidate generation unit 104 selects a partial set of attribute information from the set of attribute information as the selected attribute information based on the attribute selection evaluation information. Furthermore, the attribute candidate generation unit 104 evaluates the selected attribute information based on the attribute selection evaluation information obtained in step S204 and generates attribute candidates based on the evaluation results.

[0056] In step S206, the data model editing unit 108 obtains the attribute candidates generated by the attribute candidate generation unit 104, and the attribute information extraction information and attribute information selection evaluation information stored in the extraction setting storage unit 103. Furthermore, the data model editing unit 108 retrieves the corresponding data model from the data model information storage unit 101 based on the data model name specified in the obtained attribute information extraction information.

[0057] In step S207, the data model editing unit 108 selects evaluation information or attribute candidates based on the attribute information obtained in step S206, and generates new classification labels representing the attribute candidates. Furthermore, the data model editing unit 108 associates the generated classification labels with the attribute information contained in the attribute candidates obtained in step S206, updates the data model obtained in step S206, and thus generates a new data model.

[0058] In step S208, the data model editing unit 108 saves the data model generated in step S207 to the data model information storage unit 101, or rewrites it to the original data model stored in the data model information storage unit 101.

[0059] The above provides an overview of the data model generation apparatus. Next, we will describe in detail the structural elements of the data model generation apparatus.

[0060] <Set up a reception department>

[0061] The receiving unit 107 extracts information related to user-received attributes that improve retrieval and visibility from the vast data in the data model, and formats and stores this extracted information in the extracted information storage unit 103. The extracted information is as follows: Figure 3 The information shown is roughly divided into two categories: attribute information extraction information and attribute information selection and evaluation information.

[0062] Figure 3 The attribute information extraction information is the information used to obtain the data model or to extract a set of attribute information from the data model.

[0063] "Data model name" is the name of the data model from which attribute information is extracted. Thus, the data model that becomes the object of attribute information extraction among the data models stored in the data model information storage unit 101 is specified.

[0064] "Attribute information object" is the base information that represents the data in the data model and is treated as attribute information.

[0065] An "entity object" is the base information representing the data in a data model, used as an entity associated with attribute information. Entities are used to update the data model regarding the selected attribute information.

[0066] A "category label object" represents the baseline information in the data model's data that serves as a category label for retrieving viewpoints or cutting points within the data model. Category labels are data points or cutting points that can be used to retrieve viewpoints from the data model; a category label summarizes some attribute information.

[0067] Figure 3 The attribute information selection evaluation information is used to select a portion of the attribute information from the set of attribute information as the selected attribute information or to evaluate the selected attribute information.

[0068] "Selection method" is an indicator used to determine the priority of the attribute information to be selected. By obtaining the selection method from the attribute candidate generation unit 104 and handing it over to the selection unit 105, the selection unit 105 uses the selection method as a guideline for selecting a portion of the attribute information from the set of attribute information.

[0069] "Evaluation method" is an indicator used to evaluate a partial set of selected attribute information, i.e., the selected attribute information. The evaluation method is obtained by the attribute candidate generation unit 104 and handed over to the evaluation unit 106, whereby the evaluation unit 106 uses the evaluation method as a guideline for calculating the evaluation value of the selected attribute information.

[0070] The "benchmark value" is a reference for determining the level of the evaluation value calculated according to the "evaluation method". The attribute candidate generation unit 104 obtains the benchmark value and uses it to determine the level of the evaluation value calculated by the evaluation unit 106.

[0071] "Classification labels" are newly generated classification labels for existing data models. Classification labels are used when generating new classification labels for existing data models in the data model editing unit 108. Classification labels can also be used when selecting attribute information in the selection unit 105 or evaluating selected attribute information in the evaluation unit 106. For example, as explained in Embodiment 2, when "power optimization" is specified as a classification label, the selection unit 105 can actively select attribute information that includes "power" or "optimization" as contained in "power optimization".

[0072] "Category label structure information" is information that corresponds to "category labels" and provides detailed information about the category labels. Category label structure information can also be used when the selection unit 105 selects a portion of the attribute information set as selection attribute information or when the evaluation unit 106 evaluates the selection attribute information. For example, in Figure 3 The selection unit 105 can actively select attribute information associated with these information, as the classification label structure information is specified by terms such as "EV", "optimization", and "smart grid".

[0073] "Aggregation upper limit" is the upper limit value of the attribute information that the selection unit 105 can select as the attribute information. When evaluating the selected attribute information, the aggregation upper limit is used when the evaluation unit 106 evaluates it using the amount of violation of the constraint of the selection upper limit.

[0074] The setting receiving unit 107 can be an input device that allows users to input text or other data, or a communication device that accepts forms, applications, etc., entered from a web browser. Furthermore, the setting receiving unit 107 can also exchange information with external devices using text files or similar documents.

[0075] In this embodiment 1 as described above, the attribute information selection evaluation information received by the setting acceptance unit 107 includes category tags and category tag structure information, and therefore corresponds to the tag information associated with the category tags. Furthermore, in embodiments 1 and 2, the attribute information selection evaluation information may include at least one of category tags and category tag structure information. Additionally, in this specification, for example, at least one of A, B, C, ..., and Z refers to any one of all combinations formed by selecting one or more from the groups A, B, C, ..., and Z.

[0076] <Extract Settings Storage Section>

[0077] The setting storage unit 103 is an information storage unit that stores the information received by the setting receiving unit 107. Figure 3Such information. When the attribute information extraction management unit 102, attribute candidate generation unit 104, and data model editing unit 108 receive output instructions, the extraction setting storage unit 103 outputs the information it holds.

[0078] <Data Model Information Storage Department>

[0079] The data model information storage unit 101 maintains multiple data models of the monitored object and manages each data model according to its name. Figure 4 This is a diagram illustrating an example of the data stored in the data model information storage unit 101. Figure 4 In the example shown, the data model information storage unit 101 stores data models with names such as "academic paper," "journal," and "web report." The data format of each data model is not limited; for example, it can be text, a relational database, a graph database, etc.

[0080] <Attribute Information Extraction Management Department>

[0081] The attribute information extraction information stored in the extraction setting storage unit 103 and the data model of the extraction object stored in the data model information storage unit 101 are input into the attribute information extraction management unit 102. The attribute information extraction management unit 102 extracts a set of attribute information from the data model of the extraction object based on the attribute information extraction information and manages it.

[0082] Figure 5 This is a flowchart illustrating the processing procedure of the attribute information extraction management unit 102 according to Embodiment 1.

[0083] In step S501, the attribute information extraction management unit 102 reads the following information from the extraction setting storage unit 103: Figure 3 The attribute information shown is extracted, and the data model name of the extracted object is obtained from the attribute information extraction information.

[0084] In step S502, the attribute information extraction management unit 102 retrieves the data model specified by the data model name obtained in step S501 from the data model information storage unit 101. Figure 3 In the extracted attribute information, "academic paper" was specified as the data model name. Therefore, it is stored in the data model information storage unit 101. Figure 4 In the case of information, the attribute information extraction management unit 102 obtains such information from the data model information storage unit 101. Figure 6 The data model shown is named "Academic Paper". Figure 6The data model shown in the example is represented as a graph database, with a hierarchical structure consisting of nodes and edges, as indicated in the footnote.

[0085] In step S503, the attribute information extraction management unit 102 extracts a set of attribute information from the data model obtained in step S502. For example, the attribute information extraction management unit 102 refers to the attribute information objects, entity objects, and category label objects included in the attribute information extraction information obtained in step S501, such as... Figure 7 The diagram illustrates how the elements of the data model are applied to attributes, entities, and category labels. In Figure 7 In the example shown, attributes are the terminal nodes of the graph, entities are the parent nodes of attributes, and category labels are the parent nodes of entities. Then, the attribute information extraction management unit 102 extracts the information possessed by the nodes classified as attributes as attribute information. Figure 8 Showing from Figure 6 An example of a collection of attribute information extracted from the names of the end nodes of a data model.

[0086] In step S504, the attribute information extraction management unit 102 extracts the attribute information extracted in step S503, such as... Figure 8 The set of attribute information shown, such as Figure 9 The entity associated with each attribute information is managed together with the entity shown. Based on the above actions, the attribute information extraction management unit 102 can manage the collection of attribute information and the entities associated with each attribute information, regardless of the data format of the data model of the extracted object. Management can be performed using either storage devices or methods that persist the information in a database.

[0087] <Attribute Candidate Generation Department>

[0088] like Figure 1 As shown, the attribute candidate generation unit 104 includes a selection unit 105 and an evaluation unit 106. However, it is not limited to this; the selection unit 105 and the evaluation unit 106 may also be provided separately from the attribute candidate generation unit 104, rather than being provided in the attribute candidate generation unit 104.

[0089] The attribute candidate generation unit 104 receives a set of attribute information managed by the attribute information extraction management unit 102 and attribute information selection evaluation information obtained from the extraction setting storage unit 103. The attribute information selection evaluation information is used as criteria for selecting and evaluating the selected attribute information. Based on the evaluation values ​​of multiple selected attribute information obtained by repeatedly selecting and calculating the evaluation values ​​of the selected attribute information, the attribute candidate generation unit 104 generates attribute candidates from the multiple selected attribute information and outputs these attribute candidates to the data model editing unit 108.

[0090] Figure 10 This is a flowchart illustrating the processing procedure of the attribute candidate generation unit 104 according to Embodiment 1.

[0091] In step S1001, the attribute candidate generation unit 104 obtains the attribute information managed by the attribute information extraction management unit 102, such as... Figure 9 The set of attribute information shown. Additionally, the attribute candidate generation unit 104 obtains the following from the extraction setting storage unit 103: Figure 3 The attribute information shown is used to select evaluation information.

[0092] In step S1002, the attribute candidate generation unit 104 uses the set of attribute information obtained in step S1001 and the attribute information obtained in step S1001 to select evaluation information to create, for example, Figure 11 The parameters shown are retained. The parameters include a "view of attribute information," which is set from, for example... Figure 9 The set of attribute information shown is the set of attribute information after removing the entity. Additionally, parameters include the initialized "selected attribute information" and "evaluation value of attribute candidates" in the attribute information selection evaluation information, and such as... Figure 3 The "Attribute Information Selection Evaluation Information" is shown.

[0093] In step S1003, the attribute candidate generation unit 104 transmits the parameters generated in step S1002 to the selection unit 105, causing the selection unit 105 to select an attribute from the set of attribute information. Therefore, the selection unit 105 selects an attribute according to the "selection method" included in the parameters. Details regarding the selection process of the selection unit 105 will be described later.

[0094] In step S1004, the attribute candidate generation unit 104 obtains an attribute information selected by the selection unit 105 and appends it to the "selected attribute information" contained in the parameters. If an attribute information selected by the selection unit 105 already exists in the "selected attribute information", the attribute candidate generation unit 104 deletes the attribute information selected by the selection unit 105 from the "selected attribute information".

[0095] In step S1005, the attribute candidate generation unit 104 transmits parameters to the evaluation unit 106, causing the evaluation unit 106 to calculate the evaluation value for a subset of the attribute information included in the "selected attribute information," i.e., the evaluation value for the selected attribute information. Therefore, the evaluation unit 106 calculates the evaluation value for the selected attribute information according to the "evaluation method" included in the parameters.

[0096] In step S1006, the attribute candidate generation unit 104 obtains the evaluation value calculated by the evaluation unit 106 and uses the evaluation value to update the "evaluation value of the selection attribute information" contained in the parameters.

[0097] In step S1007, the attribute candidate generation unit 104 appropriately calculates the baseline value based on the "baseline value" included in the parameter. Furthermore, the attribute candidate generation unit 104 determines whether the evaluation value obtained in step S1006 is above the baseline value. If the evaluation value is above the baseline value, the process proceeds to step S1008; otherwise, the process returns to step S1003.

[0098] In step S1008, the attribute candidate generation unit 104 generates attribute candidates by associating a partial set of attribute information contained in the "selected attribute information" of the parameters with their corresponding entities, and outputs the generated candidates to the data model editing unit 108. For example, the set of attribute information and the entities have the following characteristics: Figure 9 In such a correspondence, when the selected attribute information above the baseline value is "EV", "optimization", or "smart grid", the attribute candidate generation unit 104 outputs as follows: Figure 12 Such attribute candidates.

[0099] By repeating steps S1003 to S1007 by the attribute candidate generation unit 104, the combination of attribute information contained in the selection attribute information is appropriately modified, and the evaluation values ​​of multiple selection attribute information are calculated within a certain time span. Then, by performing step S1008 by the attribute candidate generation unit 104, selection attribute information above a threshold among the multiple selection attribute information is generated as attribute candidates within a certain time span.

[0100] <Selection Department>

[0101] The selection unit 105 inputs parameters provided by the attribute candidate generation unit 104. The selection unit 105 selects an attribute from the set of attribute information included in the "list of attribute information" contained in the parameters and outputs it to the attribute candidate generation unit 104. This is achieved by the attribute candidate generation unit 104 performing the above-described process. Figure 10 The processing procedure, thereby selecting part 105 from the set of attribute information to select a portion of the attribute information as the selected attribute information.

[0102] Figure 13 This is a flowchart illustrating the processing procedure of the selection unit 105 according to Embodiment 1.

[0103] In step S1301, the selection unit 105 obtains the following from the attribute candidate generation unit 104: Figure 11 The parameters are shown.

[0104] In step S1302, the selection unit 105 selects an attribute from the set of attribute information set in the "List of Attribute Information" according to the "selection method" included in the parameters obtained in step S1301. In the selection method, at least one of the category label and category label structure information is specified, and attribute information associated with the specified information is preferentially selected. For example, in... Figure 11 Among the parameters shown, the "Selection Method" specifies the degree of association with the category label structure information, so the attribute information associated with the category label structure information is selected first.

[0105] exist Figure 14 In the example shown, "EV" exists in both the attribute information set and the category label structure information set in the "Attribute Information Overview". Therefore, in the example shown... Figure 11 When the "selection method" is specified, the selection unit 105 preferentially selects "EV". Furthermore, although not shown, when the association degree with the category label is specified in the "selection method", the selection unit 105 preferentially selects attribute information associated with the category label.

[0106] In step S1303, the selection unit 105 outputs the attribute information selected in step S1302 to the attribute candidate generation unit 104. Furthermore, in Figure 14 In the example shown, the attribute information set and category label structure information in the "Attribute Information Overview" both contain "EV", "Optimization", and "Smart Grid". Therefore, in Figure 10 If the processing steps S1003 to S1007 are repeated, a partial set of attribute information such as "EV", "optimization", and "smart grid" is preferentially selected as the selected attribute information.

[0107] <Evaluation Department>

[0108] The evaluation unit 106 inputs parameters provided by the attribute candidate generation unit 104. The evaluation unit 106 calculates the evaluation value of the "selection attribute information" contained in the parameters and outputs it to the attribute candidate generation unit 104.

[0109] Figure 15 This is a flowchart illustrating the processing procedure of the evaluation unit 106 according to Embodiment 1.

[0110] In step S1501, the evaluation unit 106 obtains parameters from the attribute candidate generation unit 104 that reflect the selection of the selection unit 105.

[0111] In step S1502, the evaluation unit 106 calculates the evaluation value of the "selection attribute information" according to the "evaluation method" included in the parameters obtained in step S1501. The evaluation method can be a method of calculating the evaluation value by using indicators such as the correlation between the attribute information included in the selection attribute information and the attribute information included in the parameters, or a method of calculating the evaluation value by using a weighted sum of some indicators.

[0112] For example, in Figure 11 Among the parameters shown, "relevance to the classification label structure information + violation of the upper limit of aggregation" is specified as an indicator in the "evaluation method". The evaluation unit 106 calculates the evaluation value of the selected attribute information according to this indicator. Figure 16 In the example of calculating the evaluation value shown, the evaluation unit 106 calculates the "degree of correlation with the classification label structure information" and the "limit of the aggregation limit violation" that are specified as indicators in the "evaluation method", and sums them to calculate the evaluation value.

[0113] In the former's "Relevance to Classification Label Structure Information," the terms "EV," "Optimization," and "Smart Grid" specified in the classification label structure information are included in "Selection Attribute Information," so the evaluation unit 106 calculates "3" as the relevance. In the latter's "Violation of Aggregation Limit Number Constraints," the number of attribute information included in "Selection Attribute Information" is "3," which does not violate the aggregation limit number of "3," so the evaluation unit 106 calculates "1" as the constraint violation amount. Although not illustrated, if the number of attribute information included in "Selection Attribute Information" violates the aggregation limit number, the evaluation unit 106 calculates "0" or a negative value as the constraint violation amount. Furthermore, the evaluation unit 106 sums the value calculated in the former's "Relevance to Classification Label Structure Information" and the value calculated in the latter's "Violation of Aggregation Limit Number Constraints" (in... Figure 16 In the example, “4” is calculated as the evaluation value.

[0114] In the above explanation, the evaluation unit 106 calculates the evaluation value based on the correlation degree, which is equivalent to the consistency degree between the selected attribute information and the classification label structure information, but it is not limited to this. For example, the evaluation unit 106 may also calculate the evaluation value based on the difference between the correlation degree between the selected attribute information and the classification label structure information. Furthermore, while the evaluation unit 106 calculates the evaluation value based on the correlation degree between the selected attribute information and the classification label structure information, it may also calculate the evaluation value based on the correlation degree between the selected attribute information and the classification label, or it may calculate the evaluation value based on the difference between the selected attribute information and the classification label. In other words, the evaluation unit 106 can calculate the evaluation value based on the correlation degree or the difference between the selected attribute information and the attribute information selection evaluation information.

[0115] In step S1503, the evaluation unit 106 outputs the evaluation value calculated in step S1502 to the attribute candidate generation unit 104.

[0116] <Data Model Editorial Department>

[0117] Information from the data model information storage unit 101, the attribute candidate generation unit 104, and the extraction setting storage unit 103 is input into the data model editing unit 108. The data model editing unit 108 updates the data structure of the data model based on the input information. In this embodiment 1, the data model editing unit 108 generates new classification labels by selecting evaluation information or attribute candidates based on attribute information. Furthermore, the data model editing unit 108 updates the data model by associating the generated classification labels with the attribute information contained in the attribute candidates. That is, the data model editing unit 108 updates the data model by selecting evaluation information and attribute candidates based on attribute information, or by updating the data model based on attribute candidates. The data model editing unit 108 outputs the updated data model to the data model information storage unit 101.

[0118] Figure 17 This is a flowchart illustrating the processing procedure of the data model editing unit 108 according to Embodiment 1.

[0119] In step S1701, the data model editing unit 108 obtains attribute information extraction information and attribute information selection evaluation information that are received by the user by the setting acceptance unit 107 and stored by the extraction setting storage unit 103.

[0120] In step S1702, the data model editing unit 108 obtains the data model name contained in the attribute information extraction information obtained in step S1701, and obtains the data model specified by the data model name from the data model information storage unit 101.

[0121] In step S1703, the data model editing unit 108 obtains attribute candidates from the attribute candidate generation unit 104.

[0122] In step S1704, the data model editing unit 108 generates new category labels by selecting evaluation information based on the attribute information obtained in step S1701. For example, if the attribute information selection evaluation information includes "category labels," the data model editing unit 108 uses "category labels" to generate new category labels. For example, if the attribute information selection evaluation information does not include "category labels," the data model editing unit 108 uses attribute information other than "category labels" to select evaluation information, or attribute information contained in the attribute candidates, to generate new category labels. Attribute information other than "category labels" may be, for example, category label structure information.

[0123] In step S1705, the data model editing unit 108 updates the data model by associating the attribute information contained in the attribute candidates with the classification labels generated in step S1704 through their entities. Figure 18 This is a diagram illustrating an example of updating the data model. Figure 18 The data model shown in the example is Figure 6 The data model shown is in the form of a graph database. The data model editing unit 108 connects the attribute information "EV", "Optimization", "Smart Grid" included in the attribute candidates, and the newly generated classification labels, via edges connecting the entities representing these attribute information, such as "Document A" and "Document X". Thus, as... Figure 19 As shown, the data model editing unit 108 generates a new data model that associates new category labels with attribute information contained in the attribute candidates. Furthermore, entities and newly generated category labels can be at different levels, such as in a parent-child relationship, or at the same level.

[0124] In step S1706, the data model editing unit 108 saves the data model generated in step S1705 to the data model information storage unit 101.

[0125] <Summary of Implementation Method 1>

[0126] According to the data model generation apparatus of Embodiment 1, a set of attribute information is extracted from the data model, and related attribute candidates are generated by repeatedly selecting attribute information and calculating the evaluation value of the selected attribute information. The data model is then updated based on the attribute candidates and other information. With this structure, even without detailed classification conditions or other information, a hierarchical data model with visibility and retrieval capabilities can be generated. Furthermore, when the data model editing unit 108 generates new classification labels based on attribute candidates generated by the attribute candidate generation unit 104 instead of using attribute information from external sources such as users for selection and evaluation information, a data model with unknown classification labels not anticipated by the user can be generated.

[0127] Furthermore, in this embodiment 1, the selection unit 105 selects evaluation information based on attribute information including at least one of category labels and category label structure information. With this structure, the selection of attribute information can be tailored to the user's preferences.

[0128] Furthermore, in this embodiment 1, the evaluation unit 106 calculates the evaluation value of the selected attribute information based on the correlation or difference between the selected attribute information and the classification label structure information. Based on this structure, it is possible to evaluate the selected attribute information in a way that matches the user's preferences, and even generate attribute candidates.

[0129] <Implementation Method 2>

[0130] Figure 20 This is a diagram showing the structure of the data model generation apparatus according to Embodiment 2. Hereinafter, structural elements that are the same as or similar to the above-described structural elements will be described using the same or similar reference numerals, and the different structural elements will be mainly explained.

[0131] Figure 20 Structure and Figure 1 The same structure is obtained by changing the attribute candidate generation unit 104 and the selection unit 105 of the structure to attribute candidate generation unit 104A and selection unit 105A, respectively.

[0132] As a first example, consider the following scenario: The attribute information selection evaluation information received by the setting acceptance unit 107 and stored by the extraction setting storage unit 103 contains "classification tags," but "classification tag structure information" is lacking. In this case, the attribute candidate generation unit 104A generates "classification tag structure information" based on the "classification tags" and provides parameters containing this "classification tag structure information" to the selection unit 105A. The selection unit 105A selects a portion of the attribute information from the set of attribute information as the selected attribute information based on the "classification tag structure information" contained in the parameters. Alternatively, the selection unit 105A may generate the "classification tag structure information" based on the "classification tags" instead of the attribute candidate generation unit 104A.

[0133] As a second example, consider the following situation: In the attribute information selection evaluation information received by the setting acceptance unit 107 and stored by the extraction setting storage unit 103, "category labels" and "category label structure information" are missing. In this case, the selection unit 105A selects a portion of the attribute information from the set of attribute information as the selected attribute information according to a predetermined procedure. For example, as a predetermined procedure, the selection unit 105A can select the selected attribute information randomly or using a heuristic method.

[0134] Figure 21 This is a flowchart illustrating the processing procedure of the attribute candidate generation unit 104A according to Embodiment 2. Figure 21 The processing of steps S2101, S2103~S2108 and Figure 10 The processing of steps S1001 and S1003~1008 is the same, so the following mainly describes the processing of step S2102.

[0135] In step S2102, the attribute candidate generation unit 104A uses the set of attribute information obtained in step S2101 and the attribute information obtained in step S2101 to select evaluation information to create, for example, Figure 11 The parameters shown are maintained. If a "classification label" exists in the attribute information selection evaluation information but "classification label structure information" is lacking, the attribute candidate generation unit 104A uses the "classification label" or other attribute information selection evaluation information to generate "classification label structure information." In this way, the attribute candidate generation unit 104A supplements the "classification label structure information."

[0136] Figure 22 This diagram illustrates an example of supplementary processing for "category label structure information." For instance, if "processing category" is specified in the category label, the attribute candidate generation unit 104A generates words such as "processing" or "category" from "processing category." Furthermore, the attribute candidate generation unit 104A generates "category label structure information" as parameters, containing some words that include "processing" or "category." The generated category label structure information can be in the form of words or regular expressions, etc.

[0137] In step S2103, the attribute candidate generation unit 104A transmits the parameters generated in step S2102 to the selection unit 105A, causing the selection unit 105A to select an attribute from the set of attribute information. Therefore, the selection unit 105A selects according to the "selection method" included in the parameters.

[0138] Figure 23 This is a flowchart illustrating the processing procedure of the selection unit 105A according to Embodiment 2. Figure 23 The processing of steps S2301 and S2303 and Figure 13 The processing of steps S1301 and S1303 is the same, so the following mainly describes the processing of step S2302.

[0139] In step S2302, the selection unit 105A selects an attribute from the set of attribute information set in the "List of Attribute Information" according to the "selection method" included in the parameters obtained in step S2102. In the selection method, at least one of the category label and category label structure information is specified, and attribute information associated with the specified information is preferentially selected. Furthermore, depending on the specified "selection method," the selection unit 105A can select an attribute either randomly or using a heuristic method or the like.

[0140] <Summary of Implementation Method 2>

[0141] According to the data model generation apparatus of this embodiment 2 as described above, the selection unit 105A selects selection attribute information based on the classification label structure information generated from the classification labels. With this structure, even when the "classification label structure information" is lacking, the selection unit 105A can still select selection attribute information. Furthermore, it can obtain combinations of unknown classifications, i.e., unknown attribute information, that the user has not anticipated.

[0142] Furthermore, in this embodiment 2, the selection unit 105A selects the selection attribute information according to a predetermined procedure, such as a random method or a heuristic method. With this structure, even when "category labels" and "category label structure information" are lacking, the selection unit 105A can still select the selection attribute information. Additionally, combinations of unknown categories, i.e., unknown attribute information, that the user has not anticipated can be obtained. Moreover, when using a heuristic method, compared to using a random method, it is expected that the number of trials until a data model matching the user's preferences is reduced.

[0143] <Implementation Method 3>

[0144] Figure 24 This is a diagram showing the structure of the data model generation apparatus according to Embodiment 3. Hereinafter, for structural elements in Embodiment 3 that are the same as or similar to the above-described structural elements, the same or similar reference numerals will be added, and the different structural elements will be mainly described.

[0145] Figure 24 Structure and Figure 20 The same structure is obtained by changing the attribute candidate generation unit 104A, evaluation unit 106, and setting acceptance unit 107 of the structure to attribute candidate generation unit 104B, evaluation learning unit 106B, and setting acceptance unit 107B, respectively.

[0146] The setting acceptance unit 107B has the functions of the setting acceptance unit 107 described in Embodiment 1; the function of prompting the user with at least a portion of the parameters including selection attribute information; and the function of receiving feedback from the user regarding the selection attribute information. Feedback regarding the selection attribute information may include, for example, specifying the selection attribute information that should be considered as an attribute candidate.

[0147] The evaluation and learning department 106B uses the feedback and selected attribute information provided to the user and processed by the setting and processing department 107B as learning data to learn an evaluation value. The evaluation value is calculated based on the selected attribute information and the learning result. The learning may be, for example, deep learning or other machine learning (training), the learning result may be, for example, a learned model, and the learning data may be, for example, teacher data.

[0148] Figure 25 This is a flowchart illustrating the processing procedure of the attribute candidate generation unit 104B according to Embodiment 3. Figure 25 The processing of steps S2501~S2504, S2506~S2508 and Figure 21 The processing of steps S2101~S2104 and S2106~S2108 is the same, so the following mainly describes the processing of step S2505.

[0149] In step S2505, the attribute candidate generation unit 104B transmits parameters to the evaluation learning unit 106B, causing the evaluation learning unit 106B to calculate an evaluation value for a subset of the attribute information included in the "selected attribute information". Therefore, the evaluation learning unit 106B calculates the evaluation value as follows.

[0150] First, the evaluation learning unit 106B transmits at least a portion of the parameters containing the selection attribute information to the setting receiving unit 107B. The setting receiving unit 107B then prompts the user with the at least a portion of the parameters and receives feedback from the user regarding the selection attribute information.

[0151] The evaluation learning unit 106B uses the feedback and selection attribute information received by the setting and receiving unit 107B as learning data to learn the evaluation value corresponding to the relationship between the feedback and selection attribute information. Furthermore, the evaluation learning unit 106B calculates the evaluation value of the selection attribute information based on the selection attribute information and the learning results.

[0152] Figure 26 This is a flowchart illustrating the processing procedure of the evaluation learning unit 106B according to Embodiment 3.

[0153] In step S2601, the evaluation learning unit 106B obtains parameters including "selection attribute information" from the attribute candidate generation unit 104B.

[0154] In step S2602, the evaluation learning unit 106B transmits at least a portion of the parameters obtained in step S2601 to the setting receiving unit 107B. The at least a portion of the parameters includes "selected attribute information" and at least a portion of "attribute information selection evaluation information." The at least a portion of "attribute information selection evaluation information" is information associated with the indicators specified in the "evaluation method" of the parameters, such as "classification label" and "classification label structure information."

[0155] In step S2603, the evaluation learning unit 106B obtains feedback from the setting acceptance unit 107B regarding the "selection attribute information" received by the setting acceptance unit 107B from the user.

[0156] In step S2604, the evaluation learning unit 106B uses the "selection attribute information" and the feedback obtained in step S2603 as learning data to perform learning on the evaluation value of the selection attribute information. The learning in the evaluation learning unit 106B can also be a weighted learning process that reflects the time from when feedback was obtained from the setting reception unit 107B to the evaluation value. Furthermore, the time when feedback was obtained can also be the time spent by the user inputting feedback into the setting reception unit 107B. In this case, for example, the weighting can be adjusted based on whether the user made an immediate decision about the feedback or considered it for a long time. Additionally, the time when feedback was obtained can also be the time from when the data model generation device was used until the user inputs the feedback into the setting reception unit 107B. In this case, for example, the weighting of feedback input after a shorter time since the data model generation device was used can be greater than the weighting of feedback input after a shorter time since the data model generation device was used.

[0157] Alternatively, in step S2604, the learning in the evaluation learning unit 106B can also be a weighted learning process that reflects the attributes of the user who receives feedback from the setting reception unit 107B into the evaluation value. User attributes may include, for example, at least one of the following: the user's job title, experience in the assigned role, adaptability to the assigned role, and whether they are a skilled worker. Thus, for example, the weighting of feedback from experienced users can be greater than the weighting of feedback from newcomers.

[0158] Furthermore, the evaluation learning unit 106B can also perform pre-learning using historical information that sets "selection attribute information" and feedback thereof together. Thus, after learning, the evaluation learning unit 106B can appropriately calculate the evaluation value for "selection attribute information" without requiring feedback from the user.

[0159] In step S2605, the evaluation learning unit 106B calculates an evaluation value for the "selection attribute information" contained in at least a portion of the parameters containing "selection attribute information" based on at least a portion of the parameters and the learning results obtained in step S2604.

[0160] In step S2606, the evaluation learning unit 106B outputs the evaluation value calculated in step S2605 to the attribute candidate generation unit 104B.

[0161] <Summary of Implementation Method 3>

[0162] According to the data model generation apparatus of Embodiment 3 described above, feedback and selection attribute information provided to external prompts such as users are used as learning data to learn evaluation values, and the evaluation values ​​are calculated based on the selection attribute information and the learning results. With this structure, after learning, evaluation of selection attribute information matching the user's preferences can be performed without feedback from the user.

[0163] <Variation Example>

[0164] The learning and calculation of evaluation values ​​by the evaluation learning unit 106B are not limited to the learning and calculation described in embodiment 3. For example, the evaluation learning unit 106B can also learn evaluation values ​​using previously associated category tags and attribute candidates as learning data. Furthermore, the evaluation learning unit 106B can also calculate the degree of correspondence between category tags and selected attribute information as an evaluation value based on the category tags, selected attribute information, and learning results included in the attribute information selection evaluation information received by the setting acceptance unit 107B. Even with such a structure, after learning, it is possible to evaluate selected attribute information that matches the user's preferences without requiring feedback from the user.

[0165] Furthermore, the evaluation learning unit 106B may not use previously associated classification labels and attribute candidates as learning data to learn evaluation values. For example, the evaluation learning unit 106B may transform the various attribute information of the selected attribute information into feature vectors and learn a nonlinear approximate function that makes the relationship between the distances, etc., of these vectors correspond to the level of the evaluation value, thereby learning the evaluation value.

[0166] Furthermore, in Embodiment 3, the evaluation learning unit 106B calculates an evaluation value by learning from feedback on the selection attribute information prompted to the outside, but is not limited to this. For example, if the feedback on the selection attribute information prompted to the outside is an evaluation value from the user, the evaluation unit 106 described in Embodiment 1 may also directly use the evaluation value from the user.

[0167] <Implementation Method 4>

[0168] Figure 27 This is a diagram showing the structure of the data model generation apparatus according to Embodiment 4. Hereinafter, for structural elements in Embodiment 4 that are the same as or similar to the above-described structural elements, the same or similar reference numerals will be added, and the different structural elements will be mainly described.

[0169] Figure 27 Structure and Figure 24The same structure is obtained by changing the attribute candidate generation unit 104B, selection unit 105A, evaluation learning unit 106B and setting acceptance unit 107B of the structure to attribute candidate generation unit 104C, selection learning unit 105C, evaluation learning unit 106C and setting acceptance unit 107C respectively.

[0170] The evaluation learning unit 106C has the functions of the evaluation learning unit 106B according to embodiment 3, and a loop function that feeds back the evaluation value of the selection attribute information to the selection learning unit 105C.

[0171] The setting acceptance unit 107C has the following functions: the setting acceptance unit 107B described in Embodiment 3; the function of prompting the user with the monitoring and control screen represented by attribute information; the function of accepting the user's operation on the monitoring and control screen; and the function of accepting the plant's inherent information, which includes the sequential relationship of the plant's processing steps represented by attribute information.

[0172] Select Learning Section 105C to perform at least one of the following Learning 1, Learning 2, and Learning 3. The following describes examples of performing all three learning tasks using Learning Section 105C.

[0173] <First Learning>

[0174] The selection learning unit 105C will use the past selection attribute information and the evaluation value of the selection attribute information fed back by the evaluation learning unit 106C as learning data to learn the selection for selection attribute information, and make selections for selection attribute information based on the set of attribute information and the learning results.

[0175] <Second Learning>

[0176] The Select Learning Unit 105C uses the user's operation history on the monitoring and control screen as learning data to learn the selection of selection attribute information, and makes selections for selection attribute information based on the set of attribute information and the learning results.

[0177] <Third Learning>

[0178] The Select Learning Unit 105C uses the factory's inherent information as learning data to learn how to select appropriate attribute information, and then makes selections based on the set of attribute information and the learning results.

[0179] After sufficient learning, the learning selection unit 105C and the evaluation learning unit 106C automatically select and evaluate the selected attribute information. Thus, after learning, without user feedback, they can select and evaluate the selected attribute information that matches the user's preferences, and even generate attribute candidates that match the user's preferences.

[0180] Figure 28 This is a flowchart illustrating the processing procedure of the attribute candidate generation unit 104C according to Embodiment 4. Figure 28 The processing of steps S2801, S2802, S2804~S2808 and Figure 25 The processes of steps S2501, S2502, and S2504 to S2508 are the same, so the following mainly describes the process of step S2803.

[0181] In step S2803, the attribute candidate generation unit 104C transmits the parameters generated in step S2802 to the selection learning unit 105C, enabling the selection learning unit 105C to select an attribute from the set of attribute information. The attribute candidate generation unit 104C includes information specifying whether to learn in the parameters, so that in steps S2803 and S2805, the selection learning unit 105C and the evaluation learning unit 106C can selectively perform learning.

[0182] Figure 29 This is a flowchart illustrating the processing procedure of the evaluation learning unit 106C according to Embodiment 4. Figure 29 The processing of steps S2901, S2903~S2907 and Figure 26 The processing of steps S2601, S2602~S2606 is the same, so the following mainly describes the processing of step S2902.

[0183] In step S2902, the evaluation and learning unit 106C determines whether to perform learning based on the information indicating whether to learn contained in the parameters obtained in step S2901. If it is determined that learning should be performed, the process proceeds to step S2903; if it is determined that learning should not be performed, the process proceeds to step S2906.

[0184] Following step S2906, in step S2907, the evaluation learning unit 106C outputs the evaluation value calculated in step S2906. The evaluation value output from the evaluation learning unit 106C is used in the attribute candidate generation unit 104C to update the "evaluation value of the selected attribute information" parameter, and is output to the selection learning unit 105C.

[0185] Figure 30 This is a flowchart illustrating the processing procedure of the selection learning unit 105C according to Embodiment 4. Figure 30 The processing of steps S3001 and S3006 Figure 23 The processing of steps S2301 and S2303 is the same, so the following mainly describes the processing of steps S3002 to S3005.

[0186] In step S3002, the learning unit 105C determines whether to perform learning based on the information indicating whether to learn contained in the parameters obtained in step S3001. If learning is determined to be performed, the process proceeds to step S3003; if learning is determined not to be performed, the process proceeds to step S3005.

[0187] In step S3003, in the selection learning unit 105C, as feedback from the evaluation learning unit 106C, the evaluation value of the "evaluation value of selection attribute information" included in the parameters is obtained as learning data. Additionally, when the setting receiving unit 107C receives the user's operation history and factory-specific information regarding the monitoring and control screen, the selection learning unit 105C obtains the operation history and factory-specific information from the setting receiving unit 107C as learning data.

[0188] In step S3004, the selection learning unit 105C uses the learning data obtained in step S3003 to learn about the selection of selection attribute information. Through this learning, the tendency of a subset of attribute information that is likely to increase the evaluation value of "selection attribute information" is learned.

[0189] In step S3005, the learning unit 105C selects an attribute from the set of attribute information that is considered likely to increase the evaluation value, based on the set of attribute information set in the "List of Attribute Information" parameter and the learning results obtained in step S3004.

[0190] <Summary of Implementation Method 4>

[0191] According to the data model generation apparatus of this embodiment 4 as described above, past selection attribute information and its evaluation value are used as learning data to learn how to select appropriate selection attributes, and the selection of appropriate selection attributes is performed based on the set of attribute information and the learning result. With this structure, after learning, attributes that are considered likely to increase the evaluation value can be preferentially selected without user feedback.

[0192] Furthermore, in this embodiment 4, the user's operation history on the monitoring and control screen represented by attribute information is used as learning data to learn the selection of attribute information, and the selection of attribute information is performed based on the set of attribute information and the learning results. According to this structure, operation history, as a type of domain information, can be used for learning, thus enabling professional knowledge in the operation of the monitoring and control system to be reflected in the selection of attribute information.

[0193] Furthermore, in this embodiment 4, factory-specific information represented by attribute information is used as learning data to learn the selection of attribute information, and the selection of attribute information is performed based on the set of attribute information and the learning results. According to this structure, factory-specific information, as a type of domain information, can be used for learning, thus enabling professional knowledge in the application of the monitoring and control system to be reflected in the selection of attribute information.

[0194] <Variation Example>

[0195] In Implementation 4, the selection learning unit 105C automatically selects for the selection attribute information, but is not limited thereto. For example, similar to the <First Learning Unit> described above, the selection learning unit 105C first uses past selection attribute information and the evaluation value of that selection attribute information as learning data to learn the selection for the selection attribute information, and selects candidates for the selection attribute information based on the set of attribute information and the learning results.

[0196] Subsequently, the setting reception unit 107C can also prompt the user for candidate selection attribute information selected by the selection learning unit 105C, and receive feedback from the user regarding the candidate selection attribute information. Furthermore, the selection learning unit 105C can also select the selection attribute information based on the feedback from the selection attribute information candidate received by the setting reception unit 107C. With this structure, the selection of selection attribute information matching the user's preferences can be performed semi-automatically.

[0197] <Implementation Method 5>

[0198] Figure 31 This is a diagram showing the structure of the data model generation apparatus according to Embodiment 5. Hereinafter, structural elements that are the same as or similar to the above-described structural elements will be described using the same or similar reference numerals, and the different structural elements will be mainly explained.

[0199] Figure 31 Structure and Figure 27 The structure obtained by changing the learning unit 105C to the learning unit 105D is the same.

[0200] The selection learning unit 105D has the functions of the selection units 105 and 105A according to embodiments 1 and 2, and the functions of the selection learning unit 105C according to embodiment 3. Furthermore, the selection learning unit 105D has the function of selecting selection attribute information based on the increase or decrease in the evaluation value of newly generated "selection attribute information". That is, the selection learning unit 105D can preferentially select attribute information without lowering the evaluation value of the "selection attribute information".

[0201] When selecting attribute information based on the increase or decrease of the evaluation value of "selected attribute information" in the learning unit 105D, the evaluation learning unit 106C does not perform learning, that is, from Figure 29 The processing in step S2902 proceeds to processing in step S2906, where the learning results are used to calculate the evaluation value. Then, in step S2907, the evaluation learning unit 106C outputs the evaluation value calculated in step S2906 to the selection learning unit 105D instead of to the attribute candidate generation unit 104C.

[0202] The selection learning unit 105D inputs parameters provided by the attribute candidate generation unit 104C and evaluation values ​​fed back from the evaluation learning unit 106C. Based on the parameters and evaluation values, the selection learning unit 105D selects an attribute from the "list of attribute information" included in the parameters. Alternatively, similar to Embodiment 4, if the receiving unit 107C has obtained information in the fields such as the user's operation history or factory-specific information, the field information can be input into the selection learning unit 105D.

[0203] Figure 32 This is a flowchart illustrating the processing procedure of the selection learning unit 105D according to Embodiment 5. Figure 32 The processing of steps S3201~S3204, S3207, and S3208 Figure 30 The processing of steps S3001~S3004, S3005 and S3006 is the same, so the following mainly describes the processing of steps S3205 and S3206.

[0204] In step S3205, the selection learning unit 105D, referring to the parameters obtained in step S3201, determines whether to cooperate with the evaluation learning unit 106C to perform collaborative processing for selecting attribute information. Alternatively, it can be determined that collaborative processing will occur with a certain probability. If collaborative processing is determined to occur, the process proceeds to step S3206; if collaborative processing is determined not to occur, the process proceeds to step S3207.

[0205] In step S3206, the selection learning unit 105D selects some attribute information from the set of attribute information included in the "List of Attribute Information" parameter and inputs it into the evaluation learning unit 106C. Alternatively, the selection learning unit 105D may use heuristic methods or the like to select some attribute information. The evaluation learning unit 106C does not perform the learning process described in the evaluation learning unit 106C, but instead uses some attribute information as the selected attribute information to calculate the evaluation value of the selected attribute information. Furthermore, the selection learning unit 105D selects one attribute information based on the calculated evaluation value. For example, the selection learning unit 105D investigates the increase or decrease of the evaluation value of the selected attribute information and selects the attribute information whose evaluation value is maximized.

[0206] <Summary of Implementation Method 5>

[0207] In the learning process of Embodiment 4, there is a tendency to select any selection attribute from among multiple similar selection attribute information within a certain range. Therefore, when the evaluation value of selection attribute information outside the certain range is the highest, it is sometimes impossible to obtain appropriate selection attribute information. In contrast, in this Embodiment 5, the selection learning unit 105D appropriately uses the same functions as the selection units 105 and 105A. Therefore, when selection attribute information outside the certain range, that is, selection attribute information that cannot be obtained through learning, is appropriate, it is possible to use that selection attribute information. In other words, the selection learning unit 105D can use selection attribute information such as mutations obtained through genetic algorithms, thus increasing the possibility of using truly optimal selection attribute information.

[0208] <Other variations>

[0209] The following will be the above Figure 1 The attribute information extraction management department 102, attribute candidate generation department 104, selection department 105, evaluation department 106, and data model editing department 108 are collectively referred to as "attribute information extraction management department 102, etc." The attribute information extraction management department 102, etc., through... Figure 33 The processing circuit 81 shown is used to implement this. Specifically, the processing circuit 81 includes: an attribute information extraction and management unit 102, which extracts a set of attribute information from the data model of the monitored object; a selection unit 105, which selects a portion of the attribute information from the set of attribute information as selected attribute information; an evaluation unit 106, which calculates an evaluation value for the selected attribute information; an attribute candidate generation unit 104, which generates attribute candidates based on the evaluation values ​​of multiple selected attribute information obtained by repeatedly selecting and calculating the evaluation values ​​of the selected attribute information; and a data model editing unit 108, which updates the data model based on label information associated with the classification labels representing the attribute candidates and the attribute candidates, or based on the attribute candidates. In the processing circuit 81, either dedicated hardware or a processor that executes a program stored in memory can be used. Examples of processors include central processing units, processing devices, arithmetic units, microprocessors, microcomputers, and DSPs (Digital Signal Processors).

[0210] When the processing circuit 81 is dedicated hardware, it can be, for example, a single circuit, a composite circuit, a programmable processor, a parallel-programmable processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or a combination thereof. The functions of each part, such as the attribute information extraction management unit 102, can be implemented either by circuits that distribute the processing circuitry, or by aggregating the functions of each part into a single processing circuit.

[0211] When the processing circuit 81 is a processor, the functions of the attribute information extraction management unit 102, etc., are implemented in combination with software, etc. Furthermore, the software, etc., includes, for example, software, firmware, or both. The software, etc., is described as a program and stored in memory. Figure 34As shown, the processor 82 applied to the processing circuit 81 reads and executes the program stored in the memory 83, thereby realizing the functions of each part. That is, the data model generation device has a memory 83 for storing a program that is executed as a result when executed by the processing circuit 81: a step of extracting a set of attribute information from the data model of the monitored object; a step of selecting a portion of the attribute information from the set of attribute information as selected attribute information; a step of calculating the evaluation value of the selected attribute information; a step of generating attribute candidates based on the evaluation values ​​of multiple selected attribute information obtained by repeatedly selecting the selected attribute information and calculating the evaluation value of the selected attribute information; and a step of updating the data model based on the label information associated with the classification label representing the attribute candidate and the attribute candidate, or based on the attribute candidate. In other words, this program can also be described as a program that enables the computer to execute the process and method of the attribute information extraction management unit 102, etc. Here, memory 83 may be, for example, non-volatile or volatile semiconductor memory such as RAM (Random Access Memory), ROM (Read Only Memory), flash memory, EPROM (Erasable Programmable Read Only Memory), EEPROM (Electrically Erasable Programmable Read Only Memory), HDD (Hard Disk Drive), magnetic disk, floppy disk, optical disk, high-density disk, mini disk, DVD (Digital Versatile Disc), their drive devices, or any storage medium used thereafter.

[0212] The above describes the structure in which the various functions of the attribute information extraction management unit 102 are implemented using either hardware or software. However, it is not limited to this; it is also possible to implement a portion of the attribute information extraction management unit 102 using dedicated hardware and another portion using software. For example, regarding the attribute information extraction management unit 102, its functions can be implemented using a processing circuit 81, which is dedicated hardware, and in addition, its functions can be implemented by the processing circuit 81, which is a processor 82, reading and executing a program stored in the memory 83.

[0213] As described above, the processing circuit 81 can implement the above-mentioned functions through hardware, software, or a combination thereof.

[0214] The data model generation device can consist of a single device or multiple devices combined together. Furthermore, the elements of the data model generation device can be constructed either by executing a program on a computer or by using hardware that does not execute a program. That is, processing can be performed using software or signal processing using hardware.

[0215] Furthermore, the data model generation apparatus described above can also be applied to a data model generation system having a data model generation unit and a setting receiving unit. The data model generation unit is a data model generation apparatus other than the setting receiving units 107, 107B, and 107C. The setting receiving unit has the same functions as the setting receiving units 107, 107B, and 107C and can communicate with the data model generation unit via the Internet or the like.

[0216] Furthermore, it is possible to freely combine various embodiments and variations, or to appropriately modify or omit various embodiments and variations.

[0217] The above description is illustrative in all respects and is not limiting. It should be understood that countless variations not illustrated can be conceived.

[0218] Explanation of symbols

[0219] 102: Attribute Information Extraction and Management Department; 104, 104A, 104B, 104C: Attribute Candidate Generation Department; 105, 105A: Selection Department; 105C, 105D: Selection and Learning Department; 106: Evaluation Department; 106B, 106C: Evaluation and Learning Department; 107, 107B, 107C: Setting and Acceptance Department; 108: Data Model Editing Department.

Claims

1. A data model generation apparatus, comprising: The attribute information extraction management department extracts a set of attribute information from the data model of the monitored objects; The selection unit selects a portion of the set of attribute information from the set of attribute information as the selected attribute information; The evaluation department calculates the evaluation value of the selected attribute information; The attribute candidate generation unit generates attribute candidates based on the evaluation values ​​of a plurality of selected attribute information obtained by repeatedly selecting the selected attribute information and calculating the evaluation value of the selected attribute information. as well as The data model editing department updates the data model based on the tag information associated with the category tag representing the attribute candidate and the attribute candidate, or based on the attribute candidate.

2. The data model generation apparatus according to claim 1, wherein, The data model generation device further includes a setting receiving unit, which receives label information from the outside, including at least one of the category labels and category label structure information corresponding to the category labels. The selection unit makes a selection based on the received tag information and the selection attribute information.

3. The data model generation apparatus according to claim 1, wherein, The data model generation device further includes a setting receiving unit, which receives the tag information containing the classification tags from an external source. Based on the classification tags contained in the accepted tag information, generate classification tag structure information corresponding to the classification tags. The selection unit makes a selection based on the classification label structure information and the selection attribute information.

4. The data model generation apparatus according to claim 1, wherein, The selection unit selects the selection attribute information according to a predetermined procedure.

5. The data model generation apparatus according to claim 1, wherein, The data model generation device also includes a receiving unit. The selection unit is a selection learning unit. This selection learning unit uses past selection attribute information and the evaluation value of the selection attribute information as learning data to learn the selection for the selection attribute information. It selects candidates for the selection attribute information based on the set of attribute information and the learning result. For the candidates of the selection attribute information that are prompted to the outside, it selects the selection attribute information based on the feedback received by the setting acceptance unit.

6. The data model generation apparatus according to claim 1, wherein, The selection unit is a selection learning unit. This selection learning unit uses the past selection attribute information and the evaluation value of the selection attribute information as learning data to learn the selection for the selection attribute information, and makes a selection for the selection attribute information based on the set of attribute information and the learning result.

7. The data model generation apparatus according to claim 1, wherein, The data model generation device also includes a setting acceptance unit, which accepts user operation history for the monitoring and control screen represented by the attribute information. The selection unit is a selection learning unit. This selection learning unit uses the operation history as learning data to learn the selection of the selection attribute information, and makes a selection of the selection attribute information based on the set of attribute information and the learning result.

8. The data model generation apparatus according to claim 1, wherein, The data model generation device further includes a setting acceptance unit, which accepts factory-specific information containing the sequential relationships of the factory's processing steps as represented by the attribute information. The selection unit is a selection learning unit. This selection learning unit uses the inherent information of the factory as learning data to learn the selection for the selection attribute information, and makes a selection for the selection attribute information based on the set of attribute information and the learning result.

9. The data model generation apparatus according to claim 1, wherein, The data model generation device further includes a setting receiving unit, which receives label information from the outside, including at least one of the category labels and category label structure information corresponding to the category labels. The evaluation department calculates the evaluation value based on the correlation or difference between the selected attribute information and the tag information.

10. The data model generation apparatus according to claim 1, wherein, The data model generation device also includes a setting and receiving unit, which receives feedback from the outside regarding the selected attribute information prompted to the outside. The evaluation department calculates the evaluation value based on the feedback.

11. The data model generation apparatus according to claim 1, wherein, The data model generation device further includes a setting receiving unit, which receives the tag information containing the classification tags from an external source. The evaluation unit is an evaluation learning unit. This evaluation learning unit learns the evaluation value by using the previously associated classification labels and attribute candidates, or the previously associated classification labels and selected attribute information as learning data, and calculates the evaluation value based on the classification labels, selected attribute information, and learning results contained in the accepted label information.

12. The data model generation apparatus according to claim 10, wherein, The evaluation unit is an evaluation learning unit that learns the evaluation value by using the feedback and the selection attribute information as learning data, and calculates the evaluation value based on the selection attribute information and the learning result. The evaluation department will weight the evaluation value based on the time it takes to receive the feedback from the designated receiving department.

13. The data model generation apparatus according to claim 10, wherein, The evaluation unit is an evaluation learning unit that learns the evaluation value by using the feedback and the selection attribute information as learning data, and calculates the evaluation value based on the selection attribute information and the learning result. The evaluation department performs a weighted calculation of the evaluation value, which reflects the attributes of the external user as received by the setting acceptance department.

14. The data model generation apparatus according to any one of claims 1 to 13, wherein, The data model editing department generates new category labels based on the label information or the attribute candidates, and associates the generated category labels with the attribute information contained in the attribute candidates, thereby updating the data model.

15. The data model generation apparatus according to any one of claims 1 to 14, wherein, The data model generation device also includes: The data model information storage unit maintains the data model used by the attribute information extraction management unit and the data model updated by the data model editing unit; and Extract the setting storage unit and retain the tag information used by the data model editing unit.

16. A data model generation system, comprising: Data model generation department; as well as The receiving unit is configured to communicate with the data model generation unit via the Internet. The data model generation unit includes: The attribute information extraction management department extracts a set of attribute information from the data model of the monitored objects; The selection unit selects a portion of the set of attribute information from the set of attribute information as the selected attribute information; The evaluation department calculates the evaluation value of the selected attribute information; An attribute candidate generation unit generates attribute candidates based on the evaluation values ​​of a plurality of selected attribute information obtained by repeatedly selecting the selected attribute information and calculating the evaluation values ​​of the selected attribute information; and The data model editing department updates the data model based on the tag information associated with the category tags representing the attribute candidates and the attribute candidates, or based on the attribute candidates. The data model generation department accepts the tag information.

17. A data model generation method, wherein, Extract a set of attribute information from the data model of the monitored object. From the set of attribute information, a portion of the set of attribute information is selected as the selected attribute information. Calculate the evaluation value of the selected attribute information. Based on the evaluation values ​​of multiple selection attribute information obtained by repeatedly selecting the selection attribute information and calculating the evaluation value of the selection attribute information, attribute candidates are generated according to the multiple selection attribute information. The data model is updated based on the tag information associated with the category tag representing the attribute candidate and the attribute candidate, or based on the attribute candidate.

Images