A new end-to-end embedding learning system for complete knowledge graphs

Abstract

The present invention discloses a new end-to-end embedded learning system for knowledge graphs, including an encoder and a decoder. The encoder is controlled by a software platform based on the decoder, and includes a document acquisition and transmission module, a data processing module, an information classification storage module and an operation module. The decoder includes a data extraction module, a system learning model and a control execution display module. The present invention realizes a new learning system for knowledge graphs from the knowledge graph data end to the decoder end through an encoder implemented by a dual weighted graph convolutional network and a decoder implemented by an expression tensor decomposition model. The document acquisition and transmission module and the network big data real-time data intercommunication and monitoring ensure the timeliness of learning and training of the learning system, and the updated knowledge graph or the new knowledge graph is learned in real time, thereby ensuring the advancement and effectiveness of the learning system.

Description

Technical Field

[0001] The present invention relates to the technical field of learning systems, and in particular to a new end-to-end embedded learning system for complete knowledge graphs. Background Art

[0002] The progress of science and technology and the development of social civilization are constantly creating a large amount of knowledge. Whether it is for modern people or modern intelligent equipment and robots, their learning ability needs to be higher. How to spend limited energy to learn and master enough knowledge in a limited time requires the guidance of a learning system;

[0003] Most of the existing learning systems are based on some deep learning or convolutional neural networks, which use historical data for repeated learning and training. When the knowledge content of the network is updated or new knowledge content appears, it is impossible to automatically select updates or additions from the network big data, resulting in the learning system not keeping up with the network trend and affecting its use. Therefore, the present invention proposes a new end-to-end embedded learning system with a complete knowledge graph to solve the problems existing in the prior art. Summary of the invention

[0004] In view of the above problems, the purpose of the present invention is to propose a new end-to-end embedded learning system for complete knowledge graphs. The new end-to-end embedded learning system for complete knowledge graphs trains the system learning model by cooperating the system learning models of the encoder and the decoder and obtaining the knowledge graph based on network big data, thereby realizing a new learning system for knowledge graphs from the knowledge graph data end to the decoder end. The timeliness of the learning and training of the learning system is ensured through real-time data intercommunication and monitoring with the network big data through the document collection and transmission module, and the updated knowledge graph or the new knowledge graph is learned in real time, thereby ensuring the advancement and effectiveness of the learning system.

[0005] To achieve the purpose of the present invention, the present invention is implemented through the following technical solutions: a new end-to-end embedded learning system for knowledge graph completion, including an encoder and a decoder, the encoder is controlled by a software platform based on the decoder, the encoder is implemented based on a dual weighted graph convolutional network, the decoder is implemented based on an expression tensor decomposition model, the encoder includes a document acquisition and transmission module, a data processing module, an information classification storage module and an operation module, the decoder includes a data extraction module, a system learning model and a control execution display module, the document acquisition and transmission module is based on network big data to obtain documents of the knowledge graph and transmit them, the data processing module is used to process the document data of the acquired knowledge graph, the information classification storage module is used to classify and store the processed knowledge graph data, the operation module is based on the system learning model of the decoder to implement operation control, the operation module is used to perform corresponding operation display on the knowledge graph data required to be retrieved by the learning module, the data extraction module is used to extract the corresponding knowledge graph data from the information classification storage module, the system learning model is based on the knowledge graph data in the information classification storage module. Deep learning training, the control execution display module is used to control and execute and display the data processed by the operation module.

[0006] Further improvements are: the document acquisition and transmission module includes a network connection submodule and a knowledge graph monitoring submodule, the network connection submodule is used for the encoder and the network big data to realize data intercommunication from the network end to the decoder end, and the knowledge graph monitoring submodule is used for real-time monitoring of data updates and data uploads of the knowledge graph in the network big data, and when there are updates and uploads of the knowledge graph, the updated knowledge graph and the new knowledge graph are downloaded through the network connection submodule as soon as possible.

[0007] A further improvement is that the data processing module includes a data conversion submodule and a data identification submodule. The data conversion submodule is used to convert knowledge graph document data of different formats into a system-compatible data format, and the data identification submodule is used to identify the converted knowledge graph data to obtain the knowledge field to which the knowledge graph belongs.

[0008] Further improvements are: the information classification storage module includes a first-order cache sub-module, a second-order storage sub-module and a storage terminal module. The first-order cache sub-module performs preliminary classification caching according to the field to which the knowledge graph identified by the data processing module belongs, the second-order storage sub-module stores different types of knowledge graphs in the same field identified by the data processing module, and the storage terminal module stores the knowledge graph data stored in the first-order cache sub-module and the second-order storage sub-module in partitioned blocks according to field and type.

[0009] A further improvement is that the data extraction module includes an information input submodule and an information identification submodule, the information input submodule is used to input knowledge graph information of the required field and type, and the information identification submodule identifies the input information to extract the corresponding knowledge graph data from the information classification storage module.

[0010] A further improvement is that the control execution display module includes an information processing sub-module and a display device, wherein the information processing sub-module is used to process the knowledge graph extracted by the data extraction module according to display requirements, and transmit the processed data to the display device for intuitive display, and the display requirements include intuitive knowledge architecture diagrams, dynamic demonstrations of knowledge architectures, and voice teaching of knowledge architectures.

[0011] A further improvement is that the control execution display module also includes an instruction conversion submodule and an execution control submodule. The instruction conversion submodule converts the knowledge graph data extracted by the data extraction module into corresponding execution instruction signals, and sends the instruction signals to the execution control submodule, which controls the corresponding device to perform the operation.

[0012] The beneficial effects of the present invention are as follows: the present invention cooperates with the system learning model of the encoder and the decoder and obtains the knowledge graph based on network big data to train the system learning model, thereby realizing a new learning system oriented to the knowledge graph from the knowledge graph data end to the decoder end, and ensures the timeliness of the learning and training of the learning system through real-time data intercommunication and monitoring with the network big data through the document collection and transmission module, and learns the updated knowledge graph or the new knowledge graph in real time, thereby ensuring the advancement and effectiveness of the learning system. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 This is a system architecture diagram of Example 1 of the present invention.

[0014] Figure 2 This is a system architecture diagram of Example 2 of the present invention. DETAILED DESCRIPTION

[0015] In order to deepen the understanding of the present invention, the present invention will be further described in detail below in conjunction with examples. The examples are only used to explain the present invention and do not constitute a limitation on the protection scope of the present invention.

[0016] Example 1

[0017] according to Figure 1As shown, this embodiment provides a new end-to-end embedded learning system for knowledge graphs, including an encoder and a decoder. The encoder is controlled by a software platform based on the decoder. The encoder is implemented based on a dual weighted graph convolutional network. The decoder is implemented based on an expression tensor decomposition model, realizing a new learning system for knowledge graphs from the knowledge graph data end to the decoder end. The encoder includes a document acquisition and transmission module, a data processing module, an information classification storage module and an operation module. The decoder includes a data extraction module, a system learning model and a control execution display module. The document acquisition and transmission module is based on network big data to obtain knowledge graphs. The document is used for transmitting, the data processing module is used for processing the document data of the acquired knowledge graph, the information classification storage module is used for classifying and storing the processed knowledge graph data, the operation module is based on the system learning model of the decoder to realize operation control, the operation module is used for performing corresponding operation display on the knowledge graph data required to be retrieved by the learning module, the data extraction module is used for extracting the corresponding knowledge graph data from the information classification storage module, the system learning model is based on the deep learning training of the knowledge graph data in the information classification storage module, and the control execution display module is used for controlling the execution and display of the data processed by the operation module.

[0018] The document acquisition and transmission module includes a network connection submodule and a knowledge graph monitoring submodule. The network connection submodule is used for the encoder and network big data to realize data intercommunication from the network end to the decoder end. The knowledge graph monitoring submodule is used to monitor the data update and data upload of the knowledge graph in the network big data in real time. When there is an update and upload of the knowledge graph, the updated knowledge graph and the new knowledge graph are downloaded through the network connection submodule as soon as possible. The real-time data intercommunication and monitoring between the document acquisition and transmission module and the network big data ensure the timeliness of the learning and training of the learning system, and the updated knowledge graph or the new knowledge graph is learned in real time, ensuring that the learning system keeps pace with the advanced network knowledge in real time.

[0019] The data processing module includes a data conversion submodule and a data identification submodule. The data conversion submodule is used to convert knowledge graph document data of different formats into a system-compatible data format. The data identification submodule is used to identify the converted knowledge graph data to obtain the knowledge field to which the knowledge graph belongs.

[0020] The information classification storage module includes a first-order cache submodule, a second-order storage submodule and a storage terminal module. The first-order cache submodule performs preliminary classification caching according to the field to which the knowledge graph identified by the data processing module belongs. The second-order storage submodule stores different types of knowledge graphs in the same field identified by the data processing module. The storage terminal module partitions and blocks the knowledge graph data stored in the first-order cache submodule and the second-order storage submodule according to field and type.

[0021] The data extraction module includes an information input submodule and an information identification submodule. The information input submodule is used to input knowledge graph information of required fields and types, and the information identification submodule identifies the input information to extract corresponding knowledge graph data from the information classification storage module.

[0022] The control execution display module includes an information processing submodule and a display device. The information processing submodule is used to process the knowledge graph extracted by the data extraction module according to display requirements, and transmit the processed data to the display device for intuitive display. The display requirements include intuitive knowledge architecture diagrams, dynamic demonstrations of knowledge architectures, and voice teaching of knowledge architectures.

[0023] Example 2

[0024] according to Figure 2 As shown, this embodiment provides a new end-to-end embedded learning system for knowledge graphs, including an encoder and a decoder. The encoder is controlled by a software platform based on the decoder. The encoder is implemented based on a dual weighted graph convolutional network. The decoder is implemented based on an expression tensor decomposition model, realizing a new learning system for knowledge graphs from the knowledge graph data end to the decoder end. The encoder includes a document acquisition and transmission module, a data processing module, an information classification storage module and an operation module. The decoder includes a data extraction module, a system learning model and a control execution display module. The document acquisition and transmission module obtains documents of the knowledge graph based on network big data and transmits them. The data processing module is used to process the document data of the acquired knowledge graph, the information classification storage module is used to classify and store the processed knowledge graph data, the operation module is based on the system learning model of the decoder to realize operation control, the operation module is used to perform corresponding operation display on the knowledge graph data required to be retrieved by the learning module, the data extraction module is used to extract the corresponding knowledge graph data from the information classification storage module, the system learning model is based on the deep learning training of the knowledge graph data in the information classification storage module, the system learning model is implemented based on a convolutional neural network, and the control execution and display module is used to control the execution and display of the data processed by the operation module.

[0025] The document acquisition and transmission module includes a network connection submodule and a knowledge graph monitoring submodule. The network connection submodule is used for the encoder and network big data to realize data intercommunication from the network end to the decoder end. The knowledge graph monitoring submodule is used to monitor the data update and data upload of the knowledge graph in the network big data in real time. When there is an update and upload of the knowledge graph, the updated knowledge graph and the new knowledge graph are downloaded through the network connection submodule as soon as possible. The real-time data intercommunication and monitoring between the document acquisition and transmission module and the network big data ensure the timeliness of the learning and training of the learning system, and the updated knowledge graph or the new knowledge graph is learned in real time, ensuring that the learning system keeps pace with the advanced network knowledge in real time.

[0026] The data processing module includes a data conversion submodule and a data identification submodule. The data conversion submodule is used to convert knowledge graph document data of different formats into a system-compatible data format. The data identification submodule is used to identify the converted knowledge graph data to obtain the knowledge field to which the knowledge graph belongs.

[0027] The information classification storage module includes a first-order cache submodule, a second-order storage submodule and a storage terminal module. The first-order cache submodule performs preliminary classification caching according to the field to which the knowledge graph identified by the data processing module belongs. The second-order storage submodule stores different types of knowledge graphs in the same field identified by the data processing module. The storage terminal module partitions and blocks the knowledge graph data stored in the first-order cache submodule and the second-order storage submodule according to field and type.

[0028] The data extraction module includes an information input submodule and an information identification submodule. The information input submodule is used to input knowledge graph information of required fields and types, and the information identification submodule identifies the input information to extract corresponding knowledge graph data from the information classification storage module.

[0029] The control execution display module also includes an instruction conversion submodule and an execution control submodule. The instruction conversion submodule converts the knowledge graph data extracted by the data extraction module into corresponding execution instruction signals, and sends the instruction signals to the execution control submodule, which controls the corresponding equipment to perform operations.

[0030] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A new end-to-end embedding learning system for complete knowledge graphs, characterized by: It includes an encoder and a decoder, wherein the encoder is controlled by a software platform based on the decoder, the encoder is implemented based on a dual weighted graph convolutional network, and the decoder is implemented based on an expression tensor decomposition model. The encoder includes a document acquisition and transmission module, a data processing module, an information classification storage module and an operation module, and the decoder includes a data extraction module, a system learning model and a control execution display module. The document acquisition and transmission module acquires and transmits documents of a knowledge graph based on network big data, the data processing module is used to process the document data of the acquired knowledge graph, and the information classification storage module is used to classify and store the processed knowledge graph data. The operation module implements operation control based on the system learning model of the decoder, and the operation module is used to perform corresponding operation display on the knowledge graph data required to be retrieved by the learning module, the data extraction module is used to extract corresponding knowledge graph data from the information classification storage module, the system learning model is trained based on deep learning of the knowledge graph data in the information classification storage module, and the control execution display module is used to control execution and display the data processed by the operation module; The document collection and transmission module includes a network connection submodule and a knowledge graph monitoring submodule. The network connection submodule is used for the encoder and the network big data to realize data intercommunication from the network end to the decoder end. The knowledge graph monitoring submodule is used for real-time monitoring of data updates and data additions of the knowledge graph in the network big data. When there is an update and addition of the knowledge graph, the updated knowledge graph and the new knowledge graph are downloaded through the network connection submodule as soon as possible.

2. A new end-to-end embedded learning system for complete knowledge graph according to claim 1, characterized in that: The data processing module includes a data conversion submodule and a data identification submodule. The data conversion submodule is used to convert knowledge graph document data of different formats into a system-compatible data format. The data identification submodule is used to identify the converted knowledge graph data to obtain the knowledge field to which the knowledge graph belongs.

3. A new end-to-end embedded learning system for complete knowledge graph according to claim 1, characterized in that: The information classification storage module includes a first-order cache submodule, a second-order storage submodule and a storage terminal module. The first-order cache submodule performs preliminary classification caching according to the field to which the knowledge graph identified by the data processing module belongs. The second-order storage submodule stores different types of knowledge graphs in the same field identified by the data processing module. The storage terminal module partitions and blocks the knowledge graph data stored in the first-order cache submodule and the second-order storage submodule according to field and type.

4. A new end-to-end embedded learning system for complete knowledge graph according to claim 1, characterized in that: The data extraction module includes an information input submodule and an information identification submodule. The information input submodule is used to input knowledge graph information of required fields and types, and the information identification submodule identifies the input information to extract corresponding knowledge graph data from the information classification storage module.

5. A new end-to-end embedded learning system for complete knowledge graph according to claim 1, characterized in that: The control execution display module includes an information processing submodule and a display device. The information processing submodule is used to process the knowledge graph extracted by the data extraction module according to display requirements, and transmit the processed data to the display device for intuitive display. The display requirements include intuitive knowledge architecture diagrams, dynamic demonstrations of knowledge architectures, and voice teaching of knowledge architectures.

6. A new end-to-end embedded learning system for knowledge graph completeness according to claim 1, characterized in that: The control execution display module also includes an instruction conversion submodule and an execution control submodule. The instruction conversion submodule converts the knowledge graph data extracted by the data extraction module into corresponding execution instruction signals, and sends the instruction signals to the execution control submodule, which controls the corresponding equipment to perform operations.

Images