Data analysis support device, data analysis support method, and program

Abstract

To provide a data analysis support apparatus capable of presenting a property which is related to a property of operation data of equipment, a data analysis support method, and a program.SOLUTION: A reception unit 105 receives a retrieval expression including a property name indicating a property of operation data of equipment and a value of the property indicated by the property name. An extraction unit 107 extracts a related property related to the property indicated by the property name included in the retrieval expression received by the reception unit 105, from a data warehouse 101 storing the operation data of the equipment. A display unit 108 displays the related property extracted by the extraction unit 107.SELECTED DRAWING: Figure 3

Description

Technical Field

[0001] The present disclosure relates to a data analysis support device, a data analysis support method, and a program.

Background Art

[0002] Devices and systems for analyzing operation data collected from facility equipment are known in order to create new services related to facility equipment, new functions of facility equipment, and the like. The operation data collected from facility equipment includes data related to a plurality of items such as an operation state, sensor values detected by sensors, etc., and each of the plurality of items has a name given by a designer of the facility equipment. Hereinafter, an item of operation data is referred to as a "property", and the name of the item is referred to as a "property name". <​​​​​​​​​​​​​​​​​​​​​​​ [Problems that the invention aims to solve]

[0006] However, the aforementioned prior art targets data where the types of property names assigned are limited, such as financial statements, and does not target operational data of equipment where different property names are assigned by different designers, or where existing property names are assigned to new properties generated when new models are added.

[0007] This disclosure is made in view of the above circumstances and aims to provide a data analysis support device, a data analysis support method, and a program that can present the properties of operating data of equipment and related properties. [Means for solving the problem]

[0008] To achieve the above objectives, the data analysis support device relating to this disclosure is The property name that indicates the properties of the operating data of the equipment and the property that the property name indicates property A means for accepting search expressions that include values, An extraction means for extracting related properties from a data warehouse where the operating data of the equipment is stored, which are related to the property indicated by the property name included in the search expression received by the reception means. The system includes a display means for displaying the related properties extracted by the extraction means. picture, The aforementioned related properties include properties that have the same property name as the property name included in the search expression but have a different meaning in terms of the data indicated by the property value, and properties that have a different property name than the property name included in the search expression but have the same meaning in terms of the data indicated by the property value. [Effects of the Invention]

[0009] According to this disclosure, it is possible to provide a data analysis support device, a data analysis support method, and a program that can present the properties of operating data of equipment and related properties. [Brief explanation of the drawing]

[0010] [Figure 1]Block diagram showing the relationship between the data analysis support device and equipment according to the embodiment. [Figure 2] Block diagram showing the hardware configuration of the data analysis support device according to the embodiment. [Figure 3] Block diagram showing the functional configuration of the data analysis support device according to the embodiment. [Figure 4] A diagram illustrating the operational data stored in the data lake according to this embodiment. [Figure 5] Figures for illustrating the information in the descriptive description of the embodiment. [Figure 6] Diagram illustrating a dictionary according to an embodiment. [Figure 7] Figure illustrating the first and second property groups according to the embodiment. [Figure 8] Diagram illustrating the third set of properties according to the embodiment. [Figure 9] Figure illustrating the extraction results according to the embodiment. [Figure 10] A flowchart showing the data analysis support process performed by the data analysis support device according to the embodiment. [Figure 11] A flowchart showing the extraction process performed by the data analysis support device according to the embodiment. [Figure 12] A flowchart showing the clustering process performed by the data analysis support device according to the embodiment. [Figure 13] A flowchart showing the clustering process performed by the data analysis support device according to the embodiment. [Modes for carrying out the invention]

[0011] (Embodiment) The data analysis support device 100 in FIG. 1 is a device for supporting the analysis of operation data collected from a plurality of facility devices 200. As shown in FIG. 1, the facility device 200 is communicably connected to the gateway device 300 by wire or wirelessly, the gateway device 300 is communicably connected to the data lake 400 via the Internet 600, the data lake 400 is communicably connected to the ETL (Extract Transform Load) device 500 by wire or wirelessly, and the ETL device 500 is communicably connected to the data analysis support device 100 by wire or wirelessly. The data analysis support device 100, the data lake 400, and the ETL device 500 are realized, for example, on a cloud server.

[0012] The data analysis support device 100 is a device for supporting the analysis of operation data collected from facility devices 200 installed in a plurality of properties. The data analysis support device 100 includes a data warehouse 101. The data warehouse 101 is a database that stores operation data converted into a searchable format by a search formula by the ETL device 500.

[0013] The facility device 200 is an electrical device installed in a property. The facility device 200 is, for example, an air conditioner, lighting equipment, a ventilation device, etc. The facility device 200 acquires operation data, for example, at a predetermined cycle, and transmits the acquired operation data to the data lake 400 via the gateway device 300.

[0014] The gateway device 300 is a device for connecting the facility device 200 to the Internet 600.

[0015] The data lake 400 is a storage that accumulates operation data collected from the facility device 200. The data lake 400 stores the collected operation data without processing it.

[0016] The ETL device 500 is a device that converts operational data stored in the data lake 400 into a format that can be searched using search queries. The ETL device 500 performs operations on the operational data stored in the data lake 400, such as unifying character codes, unifying units, and handling missing data, and converts it into a data format that can be searched using SQL (Structured Query Language), such as CSV (Comma Separated Value).

[0017] Next, the hardware configuration of the data analysis support device 100 will be described with reference to Figure 2.

[0018] The data analysis support device 100 includes a processor 11 that performs various processes, a main memory unit 12 used as a workspace for the processor 11, an auxiliary memory unit 13 that stores various data used in the processing of the processor 11, a communication unit 14 for communicating with external devices, an input unit 15 that acquires input information, and an output unit 16 that presents various information. The main memory unit 12, auxiliary memory unit 13, communication unit 14, input unit 15, and output unit 16 are all connected to the processor 11 via a bus 17.

[0019] The processor 11 includes a CPU (Central Processing Unit). The processor 11 realizes various functions of the data analysis support device 100 by executing programs stored in the auxiliary storage unit 13.

[0020] The main memory unit 12 includes RAM (Random Access Memory). Programs are loaded into the main memory unit 12 from the auxiliary memory unit 13. The main memory unit 12 is then used as a workspace for the processor 11.

[0021] The auxiliary storage unit 13 includes non-volatile memory, such as EEPROM (Electrically Erasable Programmable Read-Only Memory). In addition to the program, the auxiliary storage unit 13 stores various data used in the processing of the processor 11. The auxiliary storage unit 13 supplies data to be used by the processor 11 according to the instructions of the processor 11, and stores the data supplied by the processor 11.

[0022] The communication unit 14 includes a network interface circuit for communicating with an external device. The communication unit 14 receives a signal from the external device and outputs the data indicated by this signal to the processor 11. The communication unit 14 also transmits a signal indicating the data output from the processor 11 to the external device.

[0023] The input unit 15 includes input devices such as input keys and pointing devices. The input unit 15 acquires information entered by the user of the data analysis support device 100 and notifies the processor 11 of the acquired information.

[0024] The output unit 16 includes output devices such as an LCD (Liquid Crystal Display) and a speaker. The output unit 16 may also form a touchscreen integrated with the pointing device that constitutes the input unit 15. The output unit 16 presents various information to the user according to instructions from the processor 11.

[0025] Next, with reference to Figure 3, the functions of the data analysis support device 100 will be described.

[0026] Functionally, the data analysis support device 100 comprises a data warehouse 101 that stores operating data converted into a searchable format, a data description storage unit 102 that stores information describing properties, an individual dictionary 103 for personal use by a user, a shared dictionary 104 for joint use by multiple users, a reception unit 105 that accepts search expressions, a search unit 106 that searches the data warehouse 101 based on the search expression, an extraction unit 107 that extracts related properties of the property indicated by the property name included in the search expression, a display unit 108 that displays the related properties, an evaluation unit 109 that accepts evaluations of the related properties, a determination unit 110 that accepts a determination of whether the information registered in the individual dictionary 103 is suitable for registration in the shared dictionary, and a conversion unit 111 that converts the property name included in the search expression based on the individual dictionary 103 or the shared dictionary 104.

[0027] The data warehouse 101 stores operational data of the equipment 200, which has been converted into a searchable format by the ETL device 500. The data warehouse 101 is implemented by the auxiliary storage unit 13.

[0028] Figure 4 shows an example of operational data stored in the data warehouse 101. The operational data in Figure 4 includes the time the operational data was acquired by the equipment 200, information indicating the property where the equipment 200 is installed, information indicating the model of the equipment 200, and information about the properties of the operational data. The property information includes the property name and the property value. The properties in Figure 4 include properties with property names such as "Operation," "Thermistor 1," and "Thermistor 2." Hereinafter, the property value will be referred to as the "property value." For example, the first record in Figure 4 shows that at "May 10, 2022, 0:01," the property value "ON" for the property named "Operation," the property value "80" for the property named "Thermistor 1," and the property value "25" for the property named "Thermistor 2" were collected from equipment 200 of "Model A" in "Property 1."

[0029] The data description storage unit 102 in Figure 3 stores information about the property descriptions obtained from the specifications of the equipment 200. The descriptions stored in the data description storage unit 102 explain the meaning of the property values ​​and are used to calculate the similarity by the data description similarity calculation unit 107-12, which will be described later. The data description storage unit 102 is implemented by the auxiliary storage unit 13.

[0030] Figure 5 shows an example of the descriptive information stored in the data description storage unit 102. The descriptive information in Figure 5 includes information indicating the model of the equipment 200, the property name of the model's operating data, and the descriptive information written in the model's specifications for the property value of the property indicated by the property name. For example, the first record in Figure 5 indicates that the specifications for "Model A" described "Operating State" as the description for the property value of the property name "Operation".

[0031] The individual dictionary 103 and shared dictionary 104 in Figure 3 are dictionaries that store information regarding different spellings of property names in driving data. The individual dictionary 103 is a dictionary used individually by each of the multiple users who use the data analysis support device 100. The shared dictionary 104 is a dictionary used jointly by multiple users. The individual dictionary 103 and the shared dictionary 104 are implemented by the auxiliary storage unit 13.

[0032] Figure 6 shows an example of information regarding alternative spellings of property names stored in the individual dictionary 103. The individual dictionary 103 in Figure 6 includes information indicating the model of the equipment 200, the property name of the model's operating data, the descriptive description of the property value of the property indicated by the property name as written in the model's specifications, and the official property name registered by the user. For example, the first record in Figure 6 indicates that the specifications for "Model A" describe the property value of the property named "Thermistor 1" as "Discharge Temperature Detection," and that the user has registered "Discharge Temperature" as the official name for the property indicated by the property name "Thermistor 1" of "Model A." The official property name is registered based on the evaluation received from the user by the evaluation unit 109, as will be described later. The official property name is also used in the conversion of search expressions by the conversion unit 111, as will be described later.

[0033] The shared dictionary 104 has the same structure as in Figure 6. The official property names of the shared dictionary 104 are registered based on the judgment of the administrator of the data analysis support device 100 received by the determination unit 110, or based on evaluations from multiple users, as will be described later.

[0034] The reception unit 105 in Figure 3 accepts a search expression that includes the property name of the operating data of the equipment 200 and the property value of the property indicated by that property name. The reception unit 105 is realized through the cooperation of the processor 11 and the input unit 15. Note that the reception unit 105 is just one example of a reception means.

[0035] The search expression is used by the user to search for desired operating data from data warehouse 101. The search expression includes the property name and property value of the operating data. For example, the search expression is "properties where "Operation" is "Heating" and "Thermistor 1" is "10"°C or lower."

[0036] The search unit 106 searches the data warehouse 101, which stores the operating data of the equipment 200, based on the search formula received by the reception unit 105. The search unit 106 is implemented by the processor 11. Note that the search unit 106 is just one example of a search method.

[0037] For example, the search unit 106 searches the data warehouse 101, which stores the operation data shown in Figure 4, for operation data that satisfies the conditions of the search expression "properties where 'operation' is 'heating' and 'thermistor 1' is '10'°C or lower". The search unit 106 retrieves, for example, the record for "Property 3" in the third row of the operation data shown in Figure 4 as the search result.

[0038] The extraction unit 107 extracts related properties from the data warehouse that are associated with the property indicated by the property name included in the search expression. The extraction unit 107 is implemented by the processor 11. Note that the extraction unit 107 is just one example of an extraction method.

[0039] Related properties are properties that share the same name as a property in a search expression but have a different meaning in terms of the data represented by their property value, and properties that share a different name than a property in a search expression but have the same meaning in terms of the data represented by their property value.

[0040] For example, the extraction unit 107 extracts as related properties properties that have the property name "operation" but whose property value indicates a different meaning of data (for example, one indicates "power ON / OFF" while the other indicates "operation mode"), and properties that have a different property name from "operation" (for example, "power") but whose property value indicates the same meaning of data (for example, both indicate "power ON / OFF"). Alternatively, the extraction unit 107 extracts as related properties properties that have the property name "thermistor 1" but whose property value indicates a different meaning of data (for example, one indicates "discharge temperature detection" while the other indicates "outside air temperature"), and properties that have a different property name from "thermistor 1" (for example, "thermistor 4") but whose property value indicates the same meaning of data (for example, both indicate "discharge temperature").

[0041] Here, the extraction unit 107 includes a similarity calculation unit 107-1 and an adjustment unit 107-2.

[0042] The similarity calculation unit 107-1 calculates a first similarity between properties in the operational data stored in the data warehouse 101 that have been assigned property names included in the search expression, and a second similarity between properties in a group of properties that have been assigned property names included in the search expression and whose property values ​​are presumed to have the same meaning as the data they represent, and the properties in the operational data stored in the data warehouse 101. The first and second similarities are indices used to determine whether or not properties are similar to each other. Note that the similarity calculation unit 107-1 is an example of a similarity calculation means.

[0043] Furthermore, the similarity calculation unit 107-1 includes a data value similarity calculation unit 107-11 and a data description similarity calculation unit 107-12. The data value similarity calculation unit 107-11 calculates similarity using property values. The data description similarity calculation unit 107-12 calculates similarity using the description of property values ​​obtained from the specifications of the equipment 200. Note that the data value similarity calculation unit 107-11 is an example of a data value similarity calculation means, and the data description similarity calculation unit 107-12 is an example of a data description similarity calculation means.

[0044] The adjustment unit 107-2 defines the first and second similarity based on the data type of the property value and the number of operational data in the data warehouse 101 that have properties assigned property names, using a similarity of 1 or more calculated by the data value similarity calculation unit 107-11 and the data description similarity calculation unit 107-12. Similarity calculations require more computation for numerical data than string data, and the computation time increases with the number of target data. Therefore, in order to present related properties to the user in a stress-free time, the adjustment unit 107-2 defines the first and second similarity based on the data type of the property value and the number of data used for similarity calculation, using a similarity based on the property value and a similarity based on a similarity based on the description. Note that the adjustment unit 107-2 is just one example of an adjustment means.

[0045] The following describes in detail the method by which the extraction unit 107 extracts related properties.

[0046] First, we will explain the method by which the extraction unit 107 extracts properties that have the same property name as the property name included in the search expression, but whose meaning of the data indicated by the property value is different, based on the first similarity. For example, the reception unit 105 obtains operating data with the same property name as the property name included in the search expression it receives from the data warehouse 101, and generates a first property group consisting of the properties of the obtained operating data. For example, if the property names included in the search expression are "operation" and "thermistor 1", the first property group will include properties with "operation" or "thermistor 1" as their property name.

[0047] Next, the extraction unit 107 classifies the properties included in the first property group based on the data type of the property value. For example, it is assumed that there are two types of data types: string and numeric. For example, the property value of the property name "driving" is shown as a string as shown in Figure 4, so the extraction unit 107 classifies the property of the property name "driving" into the first property group "A1" of the string type. Also, the property value of the property name "thermistor1" is shown as a numeric value as shown in Figure 4, so the extraction unit 107 classifies the property of the property name "thermistor1" into the first property group "A2" of the numeric type.

[0048] Figure 7 shows the first property groups "A1" and "A2" which are classified based on the data type of the operating data properties. The property name "Operation" classified in the first property group "A1" is, for example, attached to the property indicating "Operating Status" in "Model A," but to the property indicating "Operating Mode" in "Model B." Similarly, the property name "Thermistor 1" classified in the first property group "A2" is, for example, attached to the property indicating "Discharge Temperature Detection" in "Model A," but to the property indicating "Outside Air Temperature" in "Model C." In other words, the first property group includes properties that have the same property name but different meanings in their property values. Therefore, next, the properties included in the first property group are classified into properties whose property values ​​indicate the same meaning in their data.

[0049] The extraction unit 107 clusters the properties of the first property group into clusters where the data represented by the property values ​​is estimated to have the same meaning. The set of clusters generated by clustering is called the second property group. That is, the clusters included in the second property group consist of properties that have the same property name and whose data represented by the property values ​​is estimated to have the same meaning. The extraction unit 107 performs clustering using the first similarity calculated by the similarity calculation unit 107-1.

[0050] First, the first similarity is calculated for properties whose property values ​​are of string type. For example, when calculating similarity using property values, string-type property values ​​require less computation than numeric-type property values. Therefore, the adjustment unit 107-2 defines the first similarity as the similarity calculated by the data value similarity calculation unit 107-11 using the property values. That is, the extraction unit 107 extracts related properties of properties with string-type property values ​​by performing clustering based on the similarity calculated using the property values.

[0051] The data value similarity calculation unit 107-11 retrieves property values ​​from the data warehouse 101 for properties whose names are included in the first string-type property group "A1". The adjustment unit 107-2 determines the period for which the data value similarity calculation unit 107-11 will search the operational data stored in the data warehouse 101. Hereinafter, this period will be referred to as the "first search period". The first search period is set to be shorter the more property names are included in the first property group "A1", and longer the fewer property names are included in the first property group "A1".

[0052] For example, if the number of property names is between 1 and 10, the search period is 1 year; if the number of property names is between 10 and 100, the search period is 1 month; and if the number of property names is 100 or more, the search period is 1 day. A table that associates the number of property names with the search period is stored in the data analysis support device 100, and the adjustment unit 107-2 refers to this table to determine the search period. The contents of this table are predetermined by the administrator of the data analysis support device 100.

[0053] For example, since the first property "A1" contains only one property name, the adjustment unit 107-2 determines the search period to be "1 year," and the data value similarity calculation unit 107-11 obtains the property value of the property assigned the property name "operation" by referring to the operation data for the past "1 year" stored in the data warehouse 101.

[0054] Next, the data value similarity calculation unit 107-11 generates a vector whose elements are all of the acquired property values. Hereafter, this vector will be referred to as the "property value vector".

[0055] For example, suppose the data value similarity calculation unit 107-11 obtains the following property values ​​for the property name "Operation": (ON,OFF) for model A, (Automatic,Cooling,Heating,Dehumidifying,Fan) for model B, (ON,OFF) for model C, and (Automatic,Cooling,Heating,Dehumidifying) for model D. In this case, the data value similarity calculation unit 107-11 defines the property value vector as (ON,OFF,Automatic,Cooling,Heating,Dehumidifying,Fan) and generates the following property value vectors for model A: (1,1,0,0,0,0,0) for model B, (0,0,1,1,1,1,1) for model C, (1,1,0,0,0,0,0) for model D, and (0,0,1,1,1,1,0) for model D.

[0056] The data value similarity calculation unit 107-11 calculates the similarity between property value vectors, and the extraction unit 107 performs clustering based on the calculated similarity. The similarity of property value vectors is calculated using, for example, the cosine value of the vectors. The clustering algorithm is a known algorithm such as Ward's method.

[0057] Figure 7 shows the second property group obtained by clustering. The properties of "Model A" and "Model C" with the property name "Operation" have property values ​​that indicate "ON" and "OFF," and as a result of clustering, they are classified into the second property group "B1." Similarly, the properties of "Model B" and "Model D" with the property name "Operation" have property values ​​that indicate "Automatic," "Cooling," "Heating," etc., and as a result of clustering, they are classified into the second property group "B2." In other words, the property with the property name "Operation" included in the second property group "B1" and the property with the property name "Operation" included in the second property group "B2" have the same property name, but the meaning of the data is different.

[0058] Next, the first similarity is calculated for properties whose property values ​​are of numerical type. The similarity of numerical properties is calculated using the property values ​​as time-series data. Here, since the similarity of numerical properties is computationally more complex than that of string properties, clustering is first performed based on the content of the descriptive descriptions, which require less computation, to reduce the number of properties to be compared for similarity within each cluster. For example, in Ward's method, the computational complexity increases in proportion to the square of the number of elements in a cluster, so roughly dividing the data beforehand using a clustering method with low processing load is an effective way to suppress the overall computational complexity. Therefore, the adjustment unit 107-2 defines the first similarity as including the similarity calculated by the data descriptive similarity calculation unit 107-12 and the similarity calculated by the data value similarity calculation unit 107-11. That is, the extraction unit 107 extracts related properties of properties with numerical property values ​​by performing clustering based on the similarity calculated using the descriptive descriptions and the similarity calculated using the property values.

[0059] Furthermore, even for properties with numerical property values, if the number of operational data entries in the data warehouse 101 that have a property name included in the search expression falls below a predetermined number, the adjustment unit 107-2 may define the first similarity as the similarity calculated by the data value similarity calculation unit 107-11. The predetermined value is set by the administrator of the data analysis support device 100.

[0060] First, the data description similarity calculation unit 107-12 calculates the similarity of the description of properties included in the first group of numerical properties, and the extraction unit 107 performs clustering based on the calculated similarity.

[0061] For example, the data description similarity calculation unit 107-12 refers to the data description storage unit 102 to obtain the description of the property name included in the first property group "A2", extracts independent words from the obtained description, and defines word vectors. Then, the data description similarity calculation unit 107-12 performs clustering based on the similarity between word vectors. Suppose the data description similarity calculation unit 107-12 extracts the following independent words from the description of the property name "thermistor 1": for model A (discharge, temperature, detection), for model B (compressor, discharge, temperature), for model C (outside air, temperature), and for model D (outside air, temperature). At this time, the data description similarity calculation unit 107-12 defines the word vector as (discharge, temperature, detection, compressor, outside air) and generates the following word vectors for model A (1,1,1,0,0), for model B (1,1,0,1,0), for model C (0,1,0,0,1), and for model D (0,1,0,0,1).

[0062] Next, the data description similarity calculation unit 107-12 calculates the similarity between word vectors. Furthermore, the word vectors may be weighted using a method such as tf-idf (term frequency - inverse document frequency). The extraction unit 107 performs clustering based on the calculated similarity. The extraction unit 107 may perform clustering using existing methods for calculating string similarity, such as Gestalt pattern matching, Levenshtein distance, Jaro-Winkler distance, or longest common subsequence. The clustering algorithm may be a known algorithm such as Ward's method. Hereinafter, clusters obtained by clustering based on the similarity of the description will be referred to as "subclusters".

[0063] Next, the adjustment unit 107-2 determines the period of operating data for which the data value similarity calculation unit 107-11 will search the operating data stored in the data warehouse 101. Hereinafter, this period will be referred to as the "second search period". Similar to the case of the first property group "A1", the adjustment unit 107-2 sets the second search period to be shorter the more property names included in the subcluster are, and longer the fewer property names included in the subcluster are.

[0064] For example, the adjustment unit 107-2 determines the second search period to "1 year," and the data value similarity calculation unit 107-11 obtains the property values ​​of the property names included in each subcluster by referring to the operating data for the past "1 year" stored in the data warehouse 101. The data value similarity calculation unit 107-11 then calculates the similarity of the time series data of the property values ​​for each subcluster, and the extraction unit 107 performs clustering based on the calculated similarity. The similarity of the time series data may be calculated, for example, by averaging the time series data for each model and aggregating the distance on the same time series. The clustering algorithm may be a known algorithm such as Ward's method.

[0065] Figure 7 shows the second property group obtained by clustering. The properties of "Model A" and "Model B" with the property name "Thermistor 1" have the independent words "discharge" and "temperature" in their descriptions, and as a result of clustering, they are classified into the second property group "B3". Similarly, the properties of "Model C" and "Model D" with the property name "Thermistor 1" have the independent words "outside air" and "temperature" in their descriptions, and as a result of clustering, they are classified into the second property group "B4". In other words, the property named "Thermistor 1" in the second property group "B3" and the property named "Thermistor 1" in the second property group "B4" have the same property name, but the meaning of the data is different.

[0066] Next, we will explain the method by which the extraction unit 107 extracts properties that have different property names from those included in the search expression, but whose data meaning is the same as that of the property value, based on the second similarity. For each cluster of the second property group, the extraction unit 107 searches the properties of the operational data stored in the data warehouse 101 for properties that are estimated to have the same data meaning as those of the property value. In this case, since all operational data stored in the data warehouse 101 is searched, the adjustment unit 107-2 defines the second similarity for properties whose property value data types are string and numeric, to include the similarity calculated by the data description similarity calculation unit 107-12 and the similarity calculated by the data value similarity calculation unit 107-11. That is, the extraction unit 107 extracts related properties of properties with numeric property values ​​by performing clustering based on the similarity calculated using the description and the similarity calculated using the property value.

[0067] For example, first, the adjustment unit 107-2 determines whether the property values ​​of the properties included in clusters "B1" to "B4" of the second property group are of string type. Next, the data description similarity calculation unit 107-12 defines a description for each of the second property groups "B1" and "B2" whose property values ​​were determined to be of string type by the adjustment unit 107-2, and calculates the similarity between the description descriptions of the second property groups and the description descriptions of the properties of the operational data stored in the data warehouse 101. Then, the data description similarity calculation unit 107-12 extracts properties that are above the threshold TH1. Threshold TH1, and thresholds TH2, TH3, and TH4, which will be described later, are predetermined by the administrator of the data analysis support device 100.

[0068] For example, the data description similarity calculation unit 107-12 defines the data description for each of the second property groups "B1" and "B2" as the sum of word vectors corresponding to the description descriptions of the properties included in each property group. The data description similarity calculation unit 107-12 then calculates the similarity between the word vectors of the description descriptions of the properties of the driving data included in the data warehouse 101 and the word vectors corresponding to the second property groups, and extracts properties whose similarity is greater than or equal to the threshold TH1.

[0069] Next, the data value similarity calculation unit 107-11 calculates the similarity based on property values ​​between the properties included in the second property group and the extracted properties that are equal to or greater than the threshold TH1. Then, the data value similarity calculation unit 107-11 extracts properties whose similarity is equal to or greater than the threshold TH2. The set of extracted properties is called the third property group. Here, the data value similarity calculation unit 107-11 defines and generates property value vectors and calculates the similarity based on the operation data for the first search period from the operation data stored in the data warehouse 101.

[0070] The properties included in the third property group are similar to those included in the clusters of the second property group in terms of their property value descriptions and data values. In other words, the properties included in the third property group include properties that have different property names than those included in the clusters of the second property group, but are presumed to have the same meaning as the data.

[0071] Figure 8 shows the third property groups obtained for each of the second property groups "B1" and "B2". For example, the third property group of the second property group "B1" with the property name "Operation" includes a property with the property name "Power Supply" that has the same description as the property included in the second property group "B1".

[0072] Next, the adjustment unit 107-2 determines whether the property values ​​of the properties included in the clusters "B1" to "B4" of the second property group are of numeric type. The data description similarity calculation unit 107-12 defines a description for each of the second property groups "B3" and "B4" whose property values ​​have been determined by the adjustment unit 107-2 to be of numeric type, and calculates the similarity between the description descriptions from the data warehouse 101, similar to the second property groups "B1" and "B2" which are of string type. Then, the data description similarity calculation unit 107-12 extracts properties that are greater than or equal to the threshold TH3. Here, considering that the calculation of similarity for properties with numeric type property values ​​is computationally more complex than the calculation of similarity for properties with string type property values, the threshold TH3 may be set to a value greater than the threshold TH1.

[0073] Next, the data value similarity calculation unit 107-11 calculates the similarity based on property values ​​between the extracted properties with a similarity of TH3 or higher and the properties included in the second property group. Then, the data value similarity calculation unit 107-11 extracts properties with a similarity of TH4 or higher as the third property group. Here, the data value similarity calculation unit 107-11 defines and generates property value vectors for the operational data of the second search period from the operational data stored in the data warehouse 101, and calculates the similarity.

[0074] Figure 8 shows the third property groups obtained for each of the second property groups "B3" and "B4". For example, the third property group of the second property group "B3" with property name "Thermistor 1" includes a property with property name "Thermistor 4" that has a description similar to the description of the property included in the second property group "B3".

[0075] The display unit 108 in Figure 3 displays the relevant properties extracted by the extraction unit 107. The display unit 108 is realized through the cooperation of the processor 11 and the output unit 16. Note that the display unit 108 is just one example of a display means.

[0076] For example, the display unit 108 displays the extraction results 700 in Figure 9 on the screen of the data analysis support device 100. The extraction results 700 in Figure 9 include information on related properties extracted by the extraction unit 107. Specifically, the extraction results 700 include tables 701-1 and 701-2 which show properties that are presumed to have the property name "Thermistor 1" assigned to different properties. Table 701-1 corresponds to the second property group "B3" extracted by the extraction unit 107, and table 701-2 corresponds to the second property group "B4" extracted by the extraction unit 107. In addition, the extraction results 700 in Figure 9 include tables 702-1 and 702-2 which show properties that are presumed to have the same meaning as the property name "Thermistor 1". Table 702-1 corresponds to the third property group for the second property group "B3" extracted by the extraction unit 107, and Table 702-2 corresponds to the third property group for the second property group "B4" extracted by the extraction unit 107.

[0077] Tables 701-1, 701-2, 702-1, and 702-2 include a property name redefinition field. When a user enters a property name in the property name redefinition field and clicks the register button 703, the entered property name is registered as the official property name in the user's individual dictionary 103, which will be described later.

[0078] Although not shown in the diagram, the extraction results displayed by the display unit 108 include a table showing properties that are presumed to have the property name "operation" assigned to a different property, as well as properties that are presumed to have the same meaning as the property name "operation," which is included in the search expression. The display unit 108 also displays the operation data of properties that satisfy the conditions of the search expression "properties where "operation" is "heating" and "thermistor 1" is "10"°C or lower" as a search result, along with the extraction results 700.

[0079] The evaluation unit 109 receives evaluations from the user regarding the related properties displayed by the display unit 108, and based on these evaluations, registers the related properties displayed by the display unit 108 in a dictionary. Here, the dictionary is the individual dictionary 103. The evaluation unit 109 is realized through the cooperation of the processor 11 and the input unit 15. Note that the evaluation unit 109 is just one example of an evaluation means.

[0080] The evaluation of related properties refers to, for example, whether or not to register the information of related properties extracted by the extraction unit 107 in the individual dictionary 103.

[0081] For example, the records included in tables 701-1, 701-2, 702-1, and 702-2 in Figure 9 can be selected by clicking. If the user selects any record and then selects the register button 703, the evaluation unit 109 determines that it has received an evaluation request to register the record's contents in the individual dictionary and registers the record's contents in the individual dictionary. For example, if user X selects the record in the third row of table 702-1 and then selects the register button 703, the evaluation unit 109 determines that it has received an evaluation request to register the record in the third row of table 702-1 in user X's individual dictionary 103 and registers the information "Model I", "Thermistor 4", and "Discharge Temperature" in user X's individual dictionary.

[0082] Furthermore, the evaluation unit 109 accepts evaluations of the individual dictionary 103 from users other than the user who uses the individual dictionary 103 for personal use. Evaluations from other users include, for example, whether or not the properties of a particular user's individual dictionary 103 are referenced by other users, and evaluations of the appropriateness of the property names by other users who have viewed a particular user's individual dictionary 103.

[0083] For example, the evaluation unit 109 receives reference requests from other users for properties registered in the individual dictionary 103, and registers the number of reference requests received or the number of people who referenced the property in the individual dictionary 103. For example, in the data analysis support device 100, users can perform data analysis using information from other users' individual dictionaries 103. If user Y views user X's individual dictionary 103 in the data analysis support device 100 and selects the property name "Thermistor 1" for "Model A", the evaluation unit 109 determines that a reference request has been received for the property name "Thermistor 1" for "Model A". The evaluation unit 109 then registers the number of references or the number of other users who referenced the property in the individual dictionary 103, associating it with the "Thermistor 1" record for "Model A".

[0084] Furthermore, for example, other users may input an evaluation of suitability for each record in the individual dictionary 103. The evaluation unit 109 receives an evaluation of "suitable" from user Y for the record of property name "thermistor 1" of "model A" registered in user X's individual dictionary 103, associates it with the record of property name "thermistor 1" of "model A", and registers the number of times an "suitable" evaluation has been received.

[0085] The determination unit 110 in Figure 3 receives a determination to determine whether the information registered in the individual dictionary 103 is suitable for registration in the shared dictionary 104, or determines the suitability for registration based on the evaluation received by the evaluation unit 109. Based on the determination that the information registered in the individual dictionary 103 is suitable for registration, it registers the information registered in the individual dictionary 103 to the shared dictionary 104. The determination unit 110 is realized through the cooperation of the processor 11 and the input unit 15. Note that the determination unit 110 is an example of a determination means.

[0086] For example, the determination unit 110 receives a determination from the administrator of the data analysis support device 100 to determine whether certain information registered in the individual dictionary 103 is suitable for registration in the shared dictionary 104. If the determination is that registration is appropriate, the determination unit 110 registers the information registered in the individual dictionary 103 in the shared dictionary 104. For example, if the determination unit 110 receives a determination from the administrator that registration is appropriate for a record registered in user X's individual dictionary 103 with model "Model A", property name "Thermistor 1", and description "Discharge temperature", it registers the record with model "Model A", property name "Thermistor 1", and description "Discharge temperature" in the shared dictionary 104.

[0087] For example, if the number of times a record has been referenced by the evaluation unit 109, or the number of other users who have referenced it, exceeds a threshold predetermined by the administrator, the determination unit 110 determines that the record exceeding the threshold is suitable for registration and registers the record in the shared dictionary 104. Alternatively, if the number of times a record has received a "suitable" evaluation registered by the evaluation unit 109 exceeds a threshold predetermined by the administrator, the determination unit 110 determines that the record exceeding the threshold is suitable for registration and registers the record in the shared dictionary 104.

[0088] The conversion unit 111 in Figure 3 converts the property name included in the search expression received by the reception unit 105 into the official property name stored in the dictionary, which is registered as the official name of the property. Then, the search unit 106 searches the data warehouse 101 based on the search expression that includes the property name converted by the conversion unit 111. Here, the dictionary may be either the individual dictionary 103 or the shared dictionary 104, or both the individual dictionary 103 and the shared dictionary 104. That is, the conversion unit 111 may convert the property name by referring to the individual dictionary 103, or by referring to the shared dictionary 104, or by referring to both the individual dictionary 103 and the shared dictionary 104. The conversion unit 111 is realized by the cooperation of the processor 11, the input unit 15, and the output unit 16. Note that the conversion unit 111 is an example of a conversion means.

[0089] For example, if the conversion unit 111 receives the search expression "Properties where 'Operation' is 'Heating' and 'Thermistor 1' is '10'°C or less" from the reception unit 105, it refers to the individual dictionary 103 in Figure 6 and presents the property names "Discharge Temperature" and "Outside Temperature," which are registered as official property names, to the user as options for conversion. Next, when the conversion unit 111 receives the user's selection of the property name "Discharge Temperature," it changes the property name "Thermistor 1" in the search expression to the property name "Discharge Temperature." Then, the search unit 106 searches the data warehouse 101 based on the converted search expression "Properties where 'Operation' is 'Heating' and 'Discharge Temperature' is '10'°C or less." Note that if only one official property name is registered, the conversion unit 111 may perform the conversion automatically without requiring a user selection. Alternatively, the conversion unit 111 may automatically convert the name to the official property name that has the most reference counts registered by the evaluation unit 109, the most number of other users who have referenced it, or the most times it has received an "appropriate" rating.

[0090] Next, the data analysis support process performed by the data analysis support device 100 according to this embodiment will be explained using the flowchart in Figure 10. The data analysis support process in Figure 10 starts, for example, when power is turned on to the data analysis support device 100.

[0091] The reception unit 105 determines whether it has received a search expression that includes the property name of the operating data of the equipment 200 and the value of the property indicated by that property name (step S101). If the reception unit 105 determines that it has received a search expression (step S101; YES), the conversion unit 111 converts the search expression by referring to the individual dictionary 103 (step S102), and further converts the search expression by referring to the shared dictionary 104 (step S103). On the other hand, if the reception unit 105 determines that it has not received a search expression (step S101; NO), the process proceeds to step S109.

[0092] For example, if the reception unit 105 receives the search query "Properties where 'Operation' is 'Heating' and 'Thermistor 1' is '10'°C or lower", it refers to the individual dictionary 103 in Figure 6 and converts the property name "Thermistor 1" to the officially registered property name "Discharge Temperature", then refers to the shared dictionary 104 and changes the property name "Operation" to "Operation Mode", which is officially registered as a property name in the shared dictionary 104.

[0093] The search unit 106 searches the data warehouse 101 based on the search expression that includes the property name converted by the conversion unit 111 (step S104).

[0094] For example, the search unit 106 searches the data warehouse 101 based on the search expression "Properties where the 'operating mode' is 'heating' and the 'discharge temperature' is '10'°C or less."

[0095] The extraction unit 107 performs an extraction process on the property names included in the search expression and extracts related properties from the data warehouse 101 that are associated with the property indicated by the property name included in the search expression. The extraction process performed by the extraction unit 107 will be explained using the flowchart in Figure 11.

[0096] The extraction unit 107 retrieves operational data from the data warehouse 101 that has the same property name as the property name included in the search expression, and generates a first property group consisting of the properties of the retrieved operational data (step S201). Then, the extraction unit 107 classifies the properties included in the first property group based on the data type of the property value (step S202).

[0097] For example, the extraction unit 107 generates a first property group consisting of properties whose property names are "operation" or "thermistor 1". Then, the extraction unit 107 classifies the property with the property name "operation" into a string-type first property group "A1", and the property with the property name "thermistor 1" into a numeric-type first property group "A2".

[0098] The extraction unit 107 determines whether or not a first property group of string type exists in the classified first property group (step S203). If the extraction unit 107 determines that a first property group of string type exists (step S203; YES), it performs clustering on the first property group of string type (step S204). The clustering on the first property group of string type will be explained using the flowchart in Figure 12.

[0099] The adjustment unit 107-2 determines the first search period (step S301). Next, the data value similarity calculation unit 107-11 searches for the operation data for the first search period and obtains the property values ​​of the searched operation data (step S302).

[0100] For example, the adjustment unit 107-2 determines the first search period to be "1 year," and the data value similarity calculation unit 107-11 obtains the property value of the property assigned the property name "operation" by referring to the operation data for the past "1 year" stored in the data warehouse 101.

[0101] Next, the data value similarity calculation unit 107-11 defines and generates a property value vector from the acquired property values ​​(step S303). Then, the data value similarity calculation unit 107-11 calculates the similarity between the property value vectors, and the extraction unit 107 performs clustering based on the calculated similarity to obtain a second group of properties (step S304).

[0102] For example, the data value similarity calculation unit 107-11 defines the property value vector as (ON,OFF,Automatic,Cooling,Heating,Dehumidifying,Fan) and generates the property value vectors (1,1,0,0,0,0,0) for model A, (0,0,1,1,1,1,1) for model B, (1,1,0,0,0,0,0) for model C, and (0,0,1,1,1,1,0) for model D, and calculates the similarity between the property value vectors. Then, the extraction unit 107 classifies the property named "Operation" into the second property groups "B1" and "B2" based on the similarity calculated by the data value similarity calculation unit 107-11, as shown in Figure 7.

[0103] Returning to step S205 in Figure 11, the extraction unit 107 determines whether or not a group of first properties of numeric type exists in the classified first property group (step S205). If the extraction unit 107 determines that a group of first properties of numeric type exists (step S205; YES), it performs clustering on the first property group of numeric type (step S206). The clustering on the first property group of numeric type will be explained using the flowchart in Figure 13.

[0104] The data description similarity calculation unit 107-12 calculates the similarity of the description descriptions of properties included in the first group of numerical properties, and the extraction unit 107 generates subclusters by performing clustering based on the calculated similarity (step S401).

[0105] For example, the data description similarity calculation unit 107-12 defines word vectors as (discharge, temperature, detection, compressor, outside air) from the description of property names included in the first property group "A2", and generates (1,1,1,0,0) for model A, (1,1,0,1,0) for model B, (0,1,0,0,1) for model C, and (0,1,0,0,1) for model D. The data description similarity calculation unit 107-12 then calculates the similarity between the word vectors, and the extraction unit 107 generates subclusters based on the calculated similarity.

[0106] The adjustment unit 107-2 determines the second search period (step S402). Next, the data value similarity calculation unit 107-11 searches for the operation data for the second search period and obtains the property values ​​of the searched operation data (step S403).

[0107] For example, the adjustment unit 107-2 determines the second search period to be "1 year," and the data value similarity calculation unit 107-11 obtains the property values ​​of the property names included in each subcluster by referring to the operation data for the past "1 year" stored in the data warehouse 101.

[0108] The data value similarity calculation unit 107-11 calculates the similarity of time-series data of property values ​​for each subcluster, and the extraction unit 107 performs clustering based on the calculated similarity to obtain the second property group (step S404).

[0109] For example, the data value similarity calculation unit 107-11 calculates the similarity of time-series data of property values ​​for each subcluster generated in step S401, and the extraction unit 107, based on the calculated similarity, classifies the property named "Thermistor 1" into the second property groups "B3" and "B4," as shown in Figure 7.

[0110] Returning to step S207 in Figure 11, the adjustment unit 107-2 determines whether the property value of the property included in the second property group is of string type (step S207). If the adjustment unit 107-2 determines that it is of string type (step S207; YES), the data description similarity calculation unit 107-12 calculates the similarity between the description of the second property group and the description of the properties of the operation data stored in the data warehouse 101, and the extraction unit 107 extracts properties with a similarity of TH1 or higher (step S208). Next, the data value similarity calculation unit 107-11 calculates the similarity based on the property values ​​between the properties included in the second property group and the extracted properties with a similarity of TH1 or higher, and the extraction unit 107 extracts properties with a similarity of TH2 or higher (step S209).

[0111] For example, if the adjustment unit 107-2 determines that the second property groups "B1" and "B2" are of string type, the data description similarity calculation unit 107-12 calculates the similarity between the word vector for the second property groups and the word vector of the description of the properties of the operation data contained in the data warehouse 101, and the extraction unit 107 extracts properties whose similarity is greater than or equal to the threshold TH1. Next, the data value similarity calculation unit 107-11 defines and generates property value vectors for the operation data of the first search period among the operation data stored in the data warehouse 101, calculates the similarity, and the extraction unit 107 extracts properties whose similarity is greater than or equal to the threshold TH2. In other words, the data value similarity calculation unit 107-11 extracts the properties of the third property group obtained for each of the second property groups "B1" and "B2" in Figure 8.

[0112] On the other hand, if the adjustment unit 107-2 determines that it is not a string type (step S207; NO), the adjustment unit 107-2 determines whether the property value of the property included in the second property group is of numeric type (step S210). If the adjustment unit 107-2 determines that it is of numeric type (step S210; YES), the data description similarity calculation unit 107-12 calculates the similarity between the description of the second property group and the description of the properties of the operation data stored in the data warehouse 101, and the extraction unit 107 extracts properties with a similarity of TH3 or higher (step S211). Next, the data value similarity calculation unit 107-11 calculates the similarity based on the property values ​​between the properties included in the second property group and the extracted properties with a similarity of TH3 or higher, and the extraction unit 107 extracts properties with a similarity of TH4 or higher (step S212). On the other hand, if the adjustment unit 107-2 determines that the value is not of numerical type (step S210; NO), the process in Figure 11 is terminated, and the process returns to step S106 in Figure 10.

[0113] For example, if the adjustment unit 107-2 determines that the second property groups "B3" and "B4" are of numerical type, the data description similarity calculation unit 107-12 calculates the similarity between the word vector for the second property groups and the word vector of the description of the properties of the operation data contained in the data warehouse 101, and the extraction unit 107 extracts properties whose similarity is greater than or equal to the threshold TH3. Next, the data value similarity calculation unit 107-11 defines and generates property value vectors for the operation data of the second search period among the operation data stored in the data warehouse 101, calculates the similarity, and the extraction unit 107 extracts properties whose similarity is greater than or equal to the threshold TH4. In other words, the data value similarity calculation unit 107-11 extracts the properties of the third property group obtained for each of the second property groups "B3" and "B4" in Figure 8. On the other hand, if the adjustment unit 107-2 determines that the type is not numerical, the process in Figure 11 is terminated, and the process returns to step S106 in Figure 10.

[0114] In step S106 of Figure 11, the display unit 108 displays the search results and extraction results (step S106). The evaluation unit 109 determines whether or not it has received an evaluation from the user for the extraction results displayed by the display unit 108 (step S107). If the evaluation unit 109 determines that it has received an evaluation from the user for the extraction results (step S107; YES), it registers the evaluation in the individual dictionary (step S108). On the other hand, if the evaluation unit 109 determines that it has not received an evaluation from the user for the extraction results (step S107; NO), it returns to step S101.

[0115] For example, the display unit 108 displays the extracted results 700 in Figure 9 along with the search results for the search expression on the screen of the data analysis support device 100. When user X selects the record in the third row of table 702-1 and selects the register button 703, the evaluation unit 109 determines that it has received an evaluation to register the record in the third row of table 702-1 in user X's individual dictionary 103, and registers the information of model "Model I", property name "Thermistor 4", and description "Discharge temperature" in user X's individual dictionary. On the other hand, if the evaluation unit 109 does not accept an evaluation for registration of the extracted results 700 in Figure 9, it returns to step S101.

[0116] In step S109, the determination unit 110 determines whether or not it has received a determination that the information registered in the individual dictionary 103 is suitable for registration in the shared dictionary 104 (step S109). If the determination unit 110 determines that it has received a determination that the information is suitable for registration (step S109; YES), it registers the information registered in the individual dictionary that was determined to be suitable for registration in the shared dictionary 104 (step S110). On the other hand, if the determination unit 110 determines that it has not received a determination that the information is suitable for registration (step S109; NO), it proceeds to step S111.

[0117] For example, if the determination unit 110 receives a determination from the administrator that a record with model "Model A", property name "Thermistor 1", and description "Discharge Temperature" registered in the individual dictionary 103 is suitable for registration, it registers the record with model "Model A", property name "Thermistor 1", and description "Discharge Temperature" in the shared dictionary 104. On the other hand, if the determination unit 110 does not receive a determination that the property information registered in the individual dictionary 103 is suitable for registration, it proceeds to step S111.

[0118] In step S111, the evaluation unit 109 determines whether or not it has received an evaluation from another user for the individual dictionary 103 (step S111). If the evaluation unit 109 determines that it has received an evaluation from another user for the individual dictionary 103 (step S111; YES), it registers the evaluation in association with the individual dictionary 103 (step S112). On the other hand, if the evaluation unit 109 determines that it has not received an evaluation from another user for the individual dictionary 103 (step S111; NO), it returns to step S101.

[0119] For example, the evaluation unit 109 receives an evaluation from user Y indicating "appropriate" for the record of the property name "thermistor 1" for "model A" registered in user X's individual dictionary 103, associates it with the record of the property name "thermistor 1" for "model A" in user X's individual dictionary 103, and registers the number of times an "appropriate" evaluation has been received. On the other hand, if the evaluation unit 109 determines that it has not received any evaluations from other users for the individual dictionary 103, it returns to step S101.

[0120] The determination unit 110 determines whether the evaluation of the individual dictionary 103 exceeds a predetermined threshold (step S113). If the determination unit 110 determines that the evaluation of the individual dictionary 103 exceeds a predetermined threshold (step S113; YES), it proceeds to step S110. On the other hand, if the determination unit 110 determines that the evaluation of the individual dictionary 103 does not exceed a predetermined threshold (step S113; NO), it returns to step S101.

[0121] For example, if the number of times the evaluation unit 109 has received a "suitable" rating exceeds a threshold predetermined by the administrator, the determination unit 110 determines that the record exceeding the threshold is suitable for registration and registers the record in the shared dictionary 104. On the other hand, if the determination unit 110 determines that the number of times the evaluation unit 109 has received a "suitable" rating does not exceed a threshold predetermined by the administrator, it returns to step S101.

[0122] According to this embodiment, for operational data properties of equipment where different property names are assigned depending on the assigner, or where existing property names are assigned to new properties generated in response to the addition of new models, the relevant properties can be presented to the user. In other words, the user can be presented with properties that have the same property name but different content, and properties that have different property names but the same content. This makes it possible to search for operational data of equipment with a wide variety of properties and inconsistent property names, and to perform highly accurate analysis.

[0123] Furthermore, according to this embodiment, the content of the process for extracting related properties can be changed depending on the data type of the property value. This improves the efficiency of the process for extracting related properties, and allows related properties to be presented to the user in a stress-free time.

[0124] Furthermore, according to this embodiment, the extracted related property information can be registered in individual dictionaries and shared dictionaries. This allows users and other users to utilize the presented related property information. In addition, the extracted related property information can be registered as official property names, and property names included in search expressions entered by users can be converted to the official property names registered in the individual dictionaries and shared dictionaries. This enables highly accurate searches for operating data of equipment that has properties with variations in property names.

[0125] (modified version) Although embodiments of this disclosure have been described above, various forms of modification and application are possible when implementing this disclosure.

[0126] In the above embodiment, each functional unit of the data analysis support device 100 was described as being implemented by a single device, but this is not limited to that. The functions of the data analysis support device 100 may be implemented by multiple devices. For example, the data warehouse 101, the data description storage unit 102, the individual dictionary 103, and the shared dictionary 104 may each be implemented by different devices.

[0127] Furthermore, although the above embodiment uses two data types, string and numeric, there may be three or more. For example, if the property value takes the form of a Boolean, integer, floating-point, or other data type, the method for calculating similarity may be changed based on those data types.

[0128] Furthermore, in the above embodiment, the first search period and the second search period are determined by referring to a predetermined table, but this is not limited to this. For example, the first search period and the second search period may be determined by a formula defined by the administrator.

[0129] Furthermore, although the data analysis support process is shown in the flowchart of Figure 10 of the above embodiment, the order of processing is not limited to this. For example, the process in step S102 may be executed after the process in step S103. Similarly, although the extraction process is shown in the flowchart of Figure 11, the order of processing is not limited to this. For example, the processes in steps S202 and 203 may be executed after the processes in steps S205 and S206. In addition, the process by which the adjustment unit 107-2 defines the first similarity and the second similarity can be performed at any time. For example, it may be performed after step S202, or before starting the extraction process in Figure 11.

[0130] Furthermore, by applying the operation program that defines the operation of the data analysis support device 100 according to the above embodiment to an existing personal computer or information terminal device, it is also possible to make the personal computer or information terminal device function as the data analysis support device 100 according to the embodiment.

[0131] Furthermore, the method of distributing such programs is optional. For example, they may be distributed by storing them on computer-readable storage media such as CD-ROMs (Compact Disk Read-Only Memory), DVDs (Digital Versatile Disks), or memory cards, or they may be distributed via communication networks such as the Internet.

[0132] This disclosure allows for various embodiments and modifications without departing from the broad spirit and scope of this disclosure. Furthermore, the embodiments described above are for illustrative purposes only and do not limit the scope of this disclosure. In other words, the scope of this disclosure is indicated by the claims, not by the embodiments. Various modifications made within the scope of the claims and the equivalent significance of the disclosure are considered to be within the scope of this disclosure.

[0133] The various aspects of this disclosure are summarized below as an appendix.

[0134] (Note 1) A receiving means that accepts a search expression that includes a property name indicating the properties of the operating data of equipment and the property value of the property indicated by that property name, An extraction means for extracting related properties from a data warehouse where the operating data of the equipment is stored, which are related to the property indicated by the property name included in the search expression received by the reception means. The system includes a display means for displaying the related properties extracted by the extraction means, Data analysis support device. (Note 2) The extraction means is The system includes a similarity calculation means for calculating a first similarity between properties of the operational data stored in the data warehouse that are assigned a property name included in the search expression, and a second similarity between a property of a group of properties that are assigned a property name included in the search expression and are presumed to have the same meaning as the data indicated by the property value, and a property of the operational data stored in the data warehouse. The aforementioned related properties include properties that have the same property name as the property name included in the search expression, but whose meaning of the data indicated by the property value is different, and properties that have a different property name than the property name included in the search expression, but whose meaning of the data indicated by the property value is the same. The extraction means is Based on the first similarity, properties are extracted that have the same property name as the property name included in the search expression, but the meaning of the data indicated by the property value is different. Based on the second similarity, properties are extracted that have different property names from those included in the search expression, but whose property values ​​represent the same data meaning. The data analysis support device described in Appendix 1. (Note 3) The similarity calculation means is, A data value similarity calculation means that calculates similarity using the aforementioned property values, A data description description similarity calculation means that calculates similarity using the description of the property value obtained from the specifications of the equipment, Includes an adjustment means that defines the first similarity and the second similarity based on the data type of the property value and the number of operational data in the data warehouse having the property name assigned to it, by one or more of the similarity calculated by the data value similarity calculation means and the similarity calculated by the data description similarity calculation means, The data analysis support device described in Appendix 2. (Note 4) A dictionary containing information regarding alternative spellings of property names of the driving data stored in the aforementioned data warehouse, The system includes an evaluation means that receives evaluations from a user for the related properties displayed by the display means, and registers the related properties displayed by the display means in the dictionary based on the evaluations. A data analysis support device as described in any one of the items 1 to 3 in the appendix. (Note 5) A conversion means that converts the property name included in the search expression received by the reception means into the official property name stored in the dictionary, which is registered as the official name of the property. The system includes a search means for searching the data warehouse based on a search expression that includes the property name converted by the conversion means, The display means displays the search results retrieved by the search means, along with the related properties. The data analysis support device described in Appendix 4. (Note 6) The aforementioned dictionary, Individual dictionaries used by each of multiple users for personal use, Includes a shared dictionary used jointly by the aforementioned multiple users, The system includes a determination means that receives a determination to determine whether the information registered in the individual dictionary is suitable for registration in the shared dictionary, or determines whether the information is suitable for registration based on the evaluation received by the evaluation means, and registers the information registered in the individual dictionary to the shared dictionary based on the determination that the information registered in the individual dictionary is suitable for registration. A data analysis support device as described in Appendix 4 or 5. (Note 7) The aforementioned data warehouse is provided, A data analysis support device as described in any one of the items 1 to 6 of the appendix. (Note 8) The reception method accepts a search expression that includes a property name indicating the properties of the operating data of the equipment and the value of the property indicated by that property name. The extraction means extracts related properties from the data warehouse where the operating data of the equipment is stored, which are associated with the property indicated by the property name included in the search expression received by the reception means. The display means displays the related properties extracted by the extraction means. Data analysis support methods. (Note 9) Computers, A receiving means that accepts a search expression that includes a property name indicating the properties of the operating data of equipment and the value of the property indicated by that property name. An extraction means extracts related properties from a data warehouse where the operating data of the equipment is stored, which are related to the property indicated by the property name included in the search expression received by the reception means. Display means for displaying related properties extracted by the extraction means, A program that makes it function as such. [Industrial applicability]

[0135] This disclosure provides a data analysis support device, a data analysis support method, and a program that can present the properties of operating data of equipment and related properties. [Explanation of Symbols]

[0136] 11 Processor, 12 Main memory unit, 13 Auxiliary memory unit, 14 Communication unit, 15 Input unit, 16 Output unit, 17 Bus, 100 Data analysis support device, 101 Data warehouse, 102 Data description storage unit, 103 Individual dictionary, 104 Shared dictionary, 106 Search unit, 107 Extraction unit, 107-1 Similarity calculation unit, 107-11 Data value similarity calculation unit, 107-12 Data description similarity calculation unit, 107-2 Adjustment unit, 108 Display unit, 109 Evaluation unit, 110 Judgment unit, 111 Conversion unit, 200 Equipment, 300 Gateway device, 400 Data lake, 500 ETL device, 600 Internet, 700 Extraction results, 701-1, 701-2, 702-1, 702-2 Tables, 703 Registration button.

Claims

1. A receiving means that accepts a search expression that includes a property name indicating the properties of the operating data of equipment and the property value of the property indicated by that property name, An extraction means for extracting related properties from a data warehouse where the operating data of the equipment is stored, which are related to the property indicated by the property name included in the search expression received by the reception means. The system includes a display means for displaying the related properties extracted by the extraction means, The aforementioned related properties include properties that have the same property name as the property name included in the search expression, but whose meaning of the data indicated by the property value is different, and properties that have a different property name than the property name included in the search expression, but whose meaning of the data indicated by the property value is the same. Data analysis support device.

2. The extraction means is The system includes a similarity calculation means for calculating a first similarity between properties of the operational data stored in the data warehouse that are assigned a property name included in the search expression, and a second similarity between a property of a property group which is a set of properties that are assigned a property name included in the search expression and whose property values ​​are presumed to have the same meaning as the data they represent, and a property of the operational data stored in the data warehouse. The extraction means is Based on the first similarity score, properties are extracted that have the same property name as the property name included in the search expression, but whose meaning of the data indicated by the property value is different. Based on the second similarity, properties are extracted that have different property names from those included in the search expression, but whose property values ​​represent the same data meaning. The data analysis support device according to claim 1.

3. The similarity calculation means is, A data value similarity calculation means that calculates similarity using the aforementioned property values, A data description description similarity calculation means that calculates similarity using the description of the property value obtained from the specifications of the equipment, Includes an adjustment means that defines the first similarity and the second similarity based on the data type of the property value and the number of operational data in the data warehouse that have the property name assigned to them, by one or more of the similarity calculated by the data value similarity calculation means and the similarity calculated by the data description similarity calculation means, The data analysis support device according to claim 2.

4. A dictionary containing information regarding alternative spellings of property names of the driving data stored in the aforementioned data warehouse, The system includes an evaluation means that receives evaluations from a user for the related properties displayed by the display means, and registers the related properties displayed by the display means in the dictionary based on the evaluations. The data analysis support device according to claim 1.

5. A conversion means that converts the property name included in the search expression received by the reception means into the official property name stored in the dictionary, which is registered as the official name of the property. The system includes a search means for searching the data warehouse based on a search expression that includes the property name converted by the conversion means, The display means displays the search results retrieved by the search means, along with the related properties. The data analysis support device according to claim 4.

6. The aforementioned dictionary, Individual dictionaries used by each of multiple users for personal use, Includes a shared dictionary used jointly by the aforementioned multiple users, The system includes a determination means that receives a determination to determine whether the information registered in the individual dictionary is suitable for registration in the shared dictionary, or determines whether the information is suitable for registration based on the evaluation received by the evaluation means, and registers the information registered in the individual dictionary to the shared dictionary based on the determination that the information registered in the individual dictionary is suitable for registration. The data analysis support device according to claim 4 or 5.

7. The aforementioned data warehouse is provided, The data analysis support device according to claim 1.

8. The reception method accepts a search expression that includes a property name indicating the properties of the operating data of the equipment and the property value of the property indicated by that property name. The extraction means extracts related properties from the data warehouse where the operating data of the equipment is stored, which are associated with the property indicated by the property name included in the search expression received by the reception means. The display means displays the related properties extracted by the extraction means, The aforementioned related properties include properties that have the same property name as the property name included in the search expression, but whose meaning of the data indicated by the property value is different, and properties that have a different property name than the property name included in the search expression, but whose meaning of the data indicated by the property value is the same. Data analysis support methods.

9. Computers, A receiving means that accepts a search expression that includes a property name indicating the properties of the operating data of equipment and the property value of the property indicated by that property name. An extraction means extracts related properties from a data warehouse where the operating data of the equipment is stored, which are related to the property indicated by the property name included in the search expression received by the reception means. Display means for displaying related properties extracted by the extraction means, To make it function as, The aforementioned related properties include properties that have the same property name as the property name included in the search expression, but whose meaning of the data indicated by the property value is different, and properties that have a different property name than the property name included in the search expression, but whose meaning of the data indicated by the property value is the same. program.

Images