Reinforced learning knowledge graph reasoning method and device based on graph convolutional neural network

Abstract

The invention discloses a reinforced learning knowledge graph reasoning method and device based on a graph convolutional neural network, and the method comprises the steps: representing a historical path and the multi-hop neighborhood information of an entity on the historical path based on a depth time sequence model and a graph convolutional neural network model, and the historical path and themulti-hop neighborhood information serving as the perception of an intelligent agent for the environment; on the basis of sensing the environment in each step, randomly selecting actions from selectable actions according to a strategy function, and adding a balance factor, so that an intelligent agent can automatically adjust the relationship and the importance of an entity to reasoning; giving anenvironmental feedback to the intelligent agent according to a knowledge graph representation learning algorithm with a path, and adding a soft feedback given by a knowledge graph representation learning model to the last step of reasoning; and finally, optimizing the parameters by maximizing the cumulative reward expectation, and finally obtaining an inference model. The knowledge reasoning model learned by the invention can improve the accuracy in reasoning tasks such as knowledge graph completion and the like, and has good practicability.

Description

technical field [0001] The present invention relates to the field of knowledge graph technology in the direction of natural language processing in the field of artificial intelligence. Specifically, the present invention relates to knowledge reasoning branches in knowledge graph technology, reinforcement learning and graph convolutional neural network technology in machine learning, and more specifically It relates to a graph convolutional neural network-based reinforcement learning knowledge map reasoning method and device. Background technique [0002] In recent years, with the continuous advancement of the wave of artificial intelligence dominated by deep learning, the development from perception to cognition is becoming the core trend of artificial intelligence. In today's environment of continuous development of technologies such as the Internet, the Internet of Things, and cloud computing, various applications emerge in an endless stream, resulting in massive data reso...

AI Technical Summary

Problems solved by technology

However, the existing model connects the relationship and the entity vector indiscriminately as the next step. This indiscriminate treatment will introduce noise to the agent's decision-making and cannot achieve the optimal decision-making strategy;

[0007] 3. In terms of model training, the existing model only gives feedback in the last step of reasoning, which makes the agent pay too much attention to the final entity, resulting in poor quality of the specific reasoning path

Images