Batch construction method of machine learning model based on knowledge graph and its server

Abstract

The present invention relates to a method for batch constructing a machine learning model based on a knowledge graph and a server thereof. The method includes: Step 1, receiving the instantiated information input by a user based on a first interface, and obtaining an initial knowledge graph instance graph corresponding to the instantiated information; Step 2, receiving the first information input by the user for each model node in each initial knowledge graph instance graph in a second interface, and obtaining a final knowledge graph instance graph corresponding to the initial knowledge graph instance graph; Step 3, for each model node in each final knowledge graph instance graph, based on the standard data pre-stored in a database, and according to a pre-set model construction strategy corresponding to the model node, batch constructing the model corresponding to each model node in each final knowledge graph instance graph.

Description

Technical Field

[0001] The present invention relates to the technical field of machine learning, and particularly to a method for batch constructing a machine learning model based on a knowledge graph and a server thereof. Background Art

[0002] Big data and artificial intelligence technologies are very popular technologies at present and are widely used in different fields, and the modeling technology is the most crucial part among them.

[0003] Currently, in terms of improving the modeling efficiency, the widely used technology is automatic modeling technology. The automatic modeling technology automatically completes complex data preprocessing processes, model construction processes, model evaluation and deployment, etc. Business experts only need to correctly understand the business and select the correct data. There are many automatic modeling software, among which the representative ones are AutoML of Google, Azure AutoML of Microsoft, Driverless of H2O, etc. These software have made different implementations in aspects such as automatic feature engineering, automatic data transformation, automatic model selection, and automatic model hyperparameter tuning, with the aim of reducing the modeling cost and improving the modeling efficiency. However, these technologies are all for training a single model. Each time, data needs to be selected, and only one model can be constructed at a time. If multiple models need to be constructed at one time, they can only be constructed multiple times, which cannot meet the requirement of batch construction. Summary of the Invention

[0004] (1) Technical Problems to be Solved

[0005] In view of the above-mentioned disadvantages and deficiencies of the prior art, the present invention provides a method for batch constructing a machine learning model based on a knowledge graph and a server thereof, which solves the technical problem that the existing automatic modeling technology can only complete the automatic construction of a single model, and for batch constructing models, the efficiency of batch constructing models can only be achieved by constructing models multiple times is relatively low.

[0006] (2) Technical Solutions

[0007] In order to achieve the above object, the main technical solutions adopted by the present invention include:

[0008] In the first aspect, an embodiment of the present invention provides a method for batch constructing a machine learning model based on a knowledge graph, and the method includes:

[0009] Step 1: Receive the instantiated information input by the user based on the first interface, and obtain an initial knowledge graph instance graph corresponding to the instantiated information;

[0010] The initial knowledge graph relationship graph preset is displayed in the first interface;

[0011] The pre - set initial knowledge graph relationship diagram is: the knowledge graph relationship diagram after a model node, which is added by the user as a child node of a specified node in the knowledge graph relationship diagram defined in advance according to the boiler system;

[0012] The initial knowledge graph instance diagram includes raw data nodes corresponding to the data collected in the boiler system, standard data nodes corresponding to standard data, and model nodes;

[0013] The standard data is the data obtained after the data collected in the boiler system is standardized;

[0014] Step 2: Receive the first information input by the user for each model node in each initial knowledge graph instance diagram in the second interface, and obtain the final knowledge graph instance diagram corresponding to this initial knowledge graph instance diagram;

[0015] The first information includes: information on establishing relationships with the pre - specified first - type standard data nodes and the pre - specified second - type standard data nodes in this initial knowledge graph instance diagram respectively;

[0016] The second interface is used to display each initial knowledge graph instance diagram;

[0017] Step 3: For each model node in each final knowledge graph instance diagram, based on the standard data pre - stored in the database, according to the pre - set model construction strategy corresponding to this model node, batch - construct the models corresponding to each model node in each final knowledge graph instance diagram.

[0018] Preferably, there is a first relationship between the specified node and the model node in the pre - set initial knowledge graph relationship diagram;

[0019] The first relationship is the relationship that the specified node contains the model node.

[0020] Preferably, step 2 specifically includes:

[0021] Receive the information that the user marks the input relationship between the model node in each initial knowledge graph instance diagram and the pre - specified first - type standard data node in this initial knowledge graph instance diagram respectively, and receive the information that the user marks the output relationship between the model node in each initial knowledge graph instance diagram and the pre - specified second - type standard data node in this initial knowledge graph instance diagram respectively, and obtain the final knowledge graph instance diagram.

[0022] Preferably, the standard data includes the field naming of the standard data, the unit of the standard data, and the pre - set information on the way of generating the standard data from the data collected in the boiler system.

[0023] Preferably, step 3 includes:

[0024] Step 3.1: For each model node in each final knowledge graph instance graph, based on the standard data pre-stored in the database, read from the database the standard data corresponding to the first type of standard data nodes having an input relationship with the model node and the standard data corresponding to the second type of standard data nodes having an output relationship with the model node.

[0025] Step 3.2: Perform data processing on the standard data read from the database so that the processed standard data conforms to the pre-set data law corresponding to the model node.

[0026] Step 3.3: For the processed standard data, use the pre-set modeling algorithm corresponding to the model node to construct and generate a model corresponding to the model node represented in a pre-set format, and save the model corresponding to the model node represented in the pre-set format to the model repository of the cloud file system according to the storage path corresponding to the model node.

[0027] The storage path corresponding to the model node is the same as the knowledge graph path of the corresponding designated node of the model node.

[0028] Preferably, the data processing includes: abnormal data filtering, outlier data filtering, and missing data filtering.

[0029] Preferably, the predefined format is: Json format;

[0030] Among them, the model represented in the pre-set format corresponding to the model node is to display the coefficients, degrees, intercepts in the model corresponding to the model node and the types of equipment in the boiler system corresponding to the model node.

[0031] Preferably, the modeling algorithm is a linear regression modeling algorithm.

[0032] Preferably, the batch-built models are deployed in the edge server for controlling the boiler system according to the pre-set deployment method.

[0033] On the other hand, this embodiment also provides a server for batch construction of a machine learning model based on a knowledge graph, and the server can execute the method for batch construction of a machine learning model based on a knowledge graph as described in any one of the above.

[0034] (III) Beneficial effects

[0035] The beneficial effects of the present invention are as follows: A method for batch constructing a machine learning model based on a knowledge graph and its server according to the present invention, due to adopting the final knowledge graph instance graph corresponding to the initial knowledge graph instance graph obtained from Step 1 and Step 2, and then, according to the standard data pre-stored in the database in Step 3, according to the preset model construction strategy corresponding to the model node, batch constructing the model corresponding to each model node in each final knowledge graph instance graph. Compared with the prior art, it batch-completes data reading and model construction, improving the efficiency of model construction. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a flowchart of a method for batch constructing a machine learning model based on a knowledge graph according to the present invention;

[0037] Figure 2 It is a schematic diagram of transforming the knowledge graph relationship diagram pre-defined according to the boiler system into a knowledge graph relationship diagram in an embodiment of the present invention;

[0038] Figure 3 It is the pre-set initial knowledge graph relationship diagram in the second embodiment of the present invention;

[0039] Figure 4 It is the final knowledge graph instance graph in the second embodiment of the present invention;

[0040] Figure 5 It is a schematic diagram of the process of batch constructing the model corresponding to each model node in the final knowledge graph instance graph for each model node in the second embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0041] In order to better explain the present invention for easy understanding, the present invention will be described in detail below with reference to the accompanying drawings through specific embodiments.

[0042] In order to better understand the above technical solution, the exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to enable a more clear and thorough understanding of the present invention and to be able to fully convey the scope of the present invention to those skilled in the art.

[0043] Embodiment 1

[0044] Refer to Figure 1 , a method for batch constructing a machine learning model based on a knowledge graph, the method comprising:

[0045] Step 1: Receive the instantiated information input by the user based on the first interface, and obtain an initial knowledge graph instance graph corresponding to the instantiated information.

[0046] An initial knowledge graph relationship graph set in advance is displayed in the first interface.

[0047] The pre-set initial knowledge graph relationship graph is: the knowledge graph relationship graph after the user adds a model node as a child node of the specified node in the knowledge graph relationship graph defined in advance according to the boiler system under the specified node.

[0048] The initial knowledge graph instance graph includes an original data node corresponding to the data collected in the boiler system, a standard data node corresponding to the standard data, and a model node.

[0049] The standard data is the data obtained after the data collected in the boiler system is standardized.

[0050] Step 2: Receive the first information input by the user for each model node in each initial knowledge graph instance graph in the second interface, and obtain a final knowledge graph instance graph corresponding to the initial knowledge graph instance graph.

[0051] The first information includes: information on establishing relationships with the pre-specified first type of standard data node and the pre-specified second type of standard data node in the initial knowledge graph instance graph respectively.

[0052] The second interface is used to display each initial knowledge graph instance graph.

[0053] Step 3: For each model node in each final knowledge graph instance graph, based on the standard data pre-stored in the database, batch construct the models corresponding to each model node in each final knowledge graph instance graph according to the pre-set model construction strategy corresponding to the model node.

[0054] In this embodiment, in the boiler system, the standard data can be "coal feeding amount of coal feeder", "control instruction of coal feeder", etc. For example, "coal feeding amount of coal feeder" is obtained by averaging multiple physical measurement point data "instantaneous coal feeding amount feedback value of coal feeder", and similarly, "control instruction of coal feeder" is obtained by averaging multiple physical measurement point data "belt speed measurement value of coal feeder". The standard data is stored in the cloud database.

[0055] In this embodiment, since the final knowledge graph instance graph corresponding to the initial knowledge graph instance graph obtained from Step 1 and Step 2 is adopted, and then, according to the standard data pre-stored in the database in Step 3, and according to the pre-set model construction strategy corresponding to the model node, models corresponding to each model node in each final knowledge graph instance graph are batch-constructed. Compared with the prior art, data reading and model construction are completed batch by batch, improving the efficiency of model construction.

[0056] Embodiment 2

[0057] See Figure 1 , this embodiment provides a method for batch constructing machine learning models based on a knowledge graph. The method includes:

[0058] Step S1: On the first interface, receive information for instantiating a pre-set initial knowledge graph relationship graph by a user to obtain a plurality of initial knowledge graph instance graphs.

[0059] The pre-set initial knowledge graph relationship graph is: a knowledge graph relationship graph after a model node, which is used as a child node of a specified node, is added by the user under the specified node in the knowledge graph relationship graph defined in advance according to the boiler system.

[0060] See Figure 2 , in this embodiment, the sub-node B in the knowledge graph relationship graph defined in advance according to the boiler system is used as the specified node, and after the user adds a model node under the specified node (sub-node B), an initial knowledge graph relationship graph is formed.

[0061] The initial knowledge graph instance graph includes an original data node corresponding to the data collected in the boiler system, a standard data node corresponding to the standard data, and a model node.

[0062] Assume that the pre-set initial knowledge graph relationship graph in this embodiment is Figure 3 as shown, where the user has pre-added, under the specified node (sub-node B), a model node that is used as a child node of the specified node (sub-node B).

[0063] The standard data is the data obtained after the data collected in the boiler system is subjected to standardization processing.

[0064] The standardization process in this embodiment is to process data into a predefined standard format. For example, multiple physical acquisition data "coal feeder coal supply" are defined as: with English names flow1, flow2, flow3, and the unit being kilograms per hour (kg / h). These 3 physical data are processed into a standard definition: with the English name coalQuality and the unit being tons per hour (t / h). The standardization processing method (flow1 + flow2 + flow3) / 3000 will be processed into 1 standard data.

[0065] The standard data includes the field naming of the standard data, the unit of the standard data, and the method information for generating the standard data from the data collected in the boiler system that is preset.

[0066] In the preset initial knowledge graph relationship diagram, there is a first relationship between the specified node and the model node, and the first relationship is the relationship that the specified node contains the model node.

[0067] Step S2: On the second interface, receive the information that the user establishes relationships between each model node in each initial knowledge graph instance diagram and the standard data nodes of the first type and the standard data nodes of the second type specified in advance in this initial knowledge graph instance diagram, to obtain the final knowledge graph instance diagram.

[0068] In the actual application of this embodiment, step S2 specifically includes:

[0069] See Figure 4 , on the second interface, receive the information that the user marks the relationship between each model node in each initial knowledge graph instance diagram and the standard data nodes of the first type specified in advance in this initial knowledge graph instance diagram as an input relationship, and receive the information that the user marks the relationship between each model node in each initial knowledge graph instance diagram and the standard data nodes of the second type specified in advance in this initial knowledge graph instance diagram as an output relationship, to obtain the final knowledge graph instance diagram.

[0070] In this embodiment, see Figure 4 , where the standard data nodes of the first type specified in advance include standard data node A and standard data node B in the figure, and the standard data nodes of the second type specified in advance include standard data node C.

[0071] Step S3: For each model node in each final knowledge graph instance diagram, based on the standard data prestored in the database, in accordance with the preset model construction strategy corresponding to this model node, batch construct the models corresponding to each model node in each final knowledge graph instance diagram.

[0072] See Figure 5, in the practical application of this embodiment, step S3 specifically includes:

[0073] Step S3.1: For each model node in each final knowledge graph instance graph, based on the standard data pre-stored in the database, read from the database the standard data corresponding to the first type of standard data nodes having an input relationship with the model node and the standard data corresponding to the second type of standard data nodes having an output relationship with the model node.

[0074] See Figure 5 , for Figure 4 the model nodes in the final knowledge graph instance graph in Figure 3 read from the database the standard data corresponding to the standard data nodes A and standard data nodes B (i.e., the first type of standard data nodes) having an input relationship with the model nodes in the final knowledge graph instance graph in Figure 4 and read from the database the standard data corresponding to the standard data node C (i.e., the second type of standard data nodes) having an output relationship with the model nodes in the final knowledge graph instance graph in

[0075] Step S3.2: Perform data processing on the standard data read from the database so that the processed standard data conforms to the pre-set data law corresponding to the model node.

[0076] The data processing includes: abnormal data filtering, outlier data filtering, missing data filtering.

[0077] Specifically, the data processing is to, before building the model, first perform data processing on the standard data read from the database to make it meet the modeling requirements, filter out abnormal data, outlier data, generate and process features according to business requirements, etc., so that the data meets the modeling requirements, that is, the standard data read from the database conforms to the pre-set data law.

[0078] Step 3.3: For the processed standard data, use the pre-set modeling algorithm corresponding to the model node to build and generate a model corresponding to the model node represented in the pre-set format, and save the model corresponding to the model node represented in the pre-set format to the model repository of the cloud file system according to the storage path corresponding to the model node.

[0079] The pre-defined format is: Json format.

[0080] Among them, the model corresponding to the model node represented in the pre-set format is to display the coefficients, degrees, intercepts in the model corresponding to the model node and the types of equipment in the boiler system corresponding to the model node.

[0081] The modeling algorithm can be a linear regression modeling algorithm, which can be selected according to actual needs in the specific implementation process and is not limited to the linear regression modeling algorithm. The storage path corresponding to the model node is the same as the knowledge graph path of the specified node corresponding to the model node. In this embodiment, the batch-built models are deployed in the edge server for controlling the boiler system according to a preset deployment method.

[0082] This embodiment also provides a server for batch constructing a machine learning model based on a knowledge graph, and the server can execute the method for batch constructing a machine learning model based on a knowledge graph in the first or second embodiment above.

[0083] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0084] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, and the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions.

[0085] It should be noted that in the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of components or steps not listed in the claim. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The present invention can be implemented by means of hardware including several different components and by means of a suitably programmed computer. In the claims listing several devices, several of these devices can be embodied by the same piece of hardware. The use of the words first, second, third, etc. is only for convenience of description and does not indicate any order. These words can be understood as part of the component name.

[0086] In addition, it should be noted that in the description of this specification, the descriptions of terms such as "one embodiment", "some embodiments", "embodiment", "example", "specific example" or "some examples" mean that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0087] Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications after learning the basic creative concepts. Therefore, the claims should be construed to include the preferred embodiments as well as all changes and modifications falling within the scope of the present invention.

[0088] Obviously, those skilled in the art can make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and its equivalent technologies, the present invention should also include these modifications and variations.

Claims

1. A method for batch construction of a machine learning model based on a knowledge graph, characterized in that, The method includes: Step 1: Receive the instantiated information input by the user based on the first interface, and obtain an initial knowledge graph instance graph corresponding to the instantiated information; The first interface displays a preset initial knowledge graph relationship graph; The preset initial knowledge graph relationship graph is: the knowledge graph relationship graph after the user adds a model node as a child node of the specified node in the knowledge graph relationship graph defined in advance according to the boiler system; The initial knowledge graph instance graph includes an original data node corresponding to the data collected in the boiler system, a standard data node corresponding to the standard data, and a model node; The standard data is the data obtained after the data collected in the boiler system is standardized; Step 2: Receive the first information input by the user for each model node in each initial knowledge graph instance graph in the second interface, and obtain a final knowledge graph instance graph corresponding to the initial knowledge graph instance graph; The first information includes: information on establishing relationships with the standard data nodes of the first type and the standard data nodes of the second type specified in advance in the initial knowledge graph instance graph respectively; The second interface is used to display each initial knowledge graph instance graph; Step 3: For each model node in each final knowledge graph instance graph, based on the standard data pre-stored in the database, batch construct the models corresponding to each model node in each final knowledge graph instance graph according to the preset model construction strategy corresponding to the model node.

2. The method for batch constructing a machine learning model based on a knowledge graph according to claim 1, wherein There is a first relationship between the specified node and the model node in the preset initial knowledge graph relationship graph; the first relationship is the relationship that the specified node contains the model node.

3. The method for batch construction of a machine learning model based on a knowledge graph according to claim 2, characterized in that, The specific content of step 2 includes: Receive the information that the user marks the model nodes in each initial knowledge graph instance graph as input relationships with the standard data nodes of the first type specified in the initial knowledge graph instance graph respectively, and receive the information that the user marks the model nodes in each initial knowledge graph instance graph as output relationships with the standard data nodes of the second type specified in the initial knowledge graph instance graph respectively, and obtain the final knowledge graph instance graph.

4. The method for batch constructing a machine learning model based on a knowledge graph according to claim 3, wherein The standard data includes the field naming of the standard data, the unit of the standard data, and the information on the preset method for generating the standard data from the data collected in the boiler system.

5. The method for batch construction of a machine learning model based on a knowledge graph according to claim 4, wherein The content of step 3 includes: Step 3.1: For each model node in each final knowledge graph instance graph, based on the standard data pre-stored in the database, read from the database the standard data corresponding to the standard data nodes of the first type that have an input relationship with the model node and the standard data corresponding to the standard data nodes of the second type that have an output relationship with the model node; Step 3.2: Perform data processing on the standard data read from the database so that the processed standard data conforms to the preset data pattern corresponding to the model node; Step 3.3: For the processed standard data, use the preset modeling algorithm corresponding to the model node to construct and generate a model corresponding to the model node represented in a preset format, and save the model corresponding to the model node represented in the preset format to the model repository of the cloud file system according to the storage path corresponding to the model node; The storage path corresponding to the model node is consistent with the knowledge graph path of the designated node corresponding to the model node.

6. The method for batch construction of a machine learning model based on a knowledge graph according to claim 5, wherein: The data processing includes: abnormal data filtering, outlier data filtering, and missing data filtering.

7. The method for batch construction of a machine learning model based on a knowledge graph according to claim 6, wherein: The predefined format is: Json format; Among them, the model represented in the preset format corresponding to the model node is to display the coefficients, degrees, intercepts in the model corresponding to the model node, and the types of equipment in the boiler system corresponding to the model node.

8. The method for batch construction of a machine learning model based on a knowledge graph according to claim 7, wherein The modeling algorithm is a linear regression modeling algorithm.

9. The method for batch construction of a machine learning model based on a knowledge graph according to claim 8, wherein: Deploy the batch-constructed model in the edge server for controlling the boiler system according to the preset deployment method.

10. A server for batch construction of a machine learning model based on a knowledge graph, characterized in that, The server can execute the method for batch construction of a machine learning model based on a knowledge graph according to any one of claims 1-9.

Images