A Revision Approach Based On Cognitive Computing For Large-Scale Knowledge Graph

Owing to the coming of cognitive computing,all walks of life have stepped on the road of intelligent transformation and upgrading.Cognitive computing aims at building a cognitive system with human cognitive function through bionic human brain's cognitive process.However,to achieve the goal of intelligentization,there is a new requirement for Machine Cognition.In the environment of large data,Knowledge Graph which is important for knowledge representation makes machine cognition possible by revealing the wholeness and relevance of human cognition from the deep level with strong semantic ability.However,because of the shortage of traditional Knowledge Graph technology and the changing of human cognition,the scale and complexity of knowledge far exceeds the cognitive ability of human beings.Investigation and research show that the errors in large data concentration are not only common but also serious,and the error rate is generally about 15%or even higher.It is difficult to reduce the error rate of the data unless there is a special institution to make strict measures to avoid errors.However,the traditional revision approach can only find and revise the simple contradiction.For the contradiction existing in the large-scale knowledge graph,the traditional logic based approach cannot cope with it.In view of the above characteristics and problems.This paper propose a revision approach for large-scale knowledge graph based on cognitive computing,which mainly includes:knowledge inference algorithm based on BP(back propagation)neural network and contradiction revision approach based on AGM belief revision.And it is verified by experiments finally.The results reveals that the revision approach which is proposed in this paper has a significant improvement in the accuracy of knowledge inference and revision.It is concluded that the revision approach is more effective when the content of the contradictory information is larger.In the end,the influence of the depth of revision subgraph on the revision effect is analyzed.The results show that the deeper the revision subgraph is,the better the revision effect of contradiction revision approach is.