The Establishment And Application Of Middle School Mathematical Knowledge Graph In Smart Education

In traditional mathematics education,students lack focus points and initiative in their learn-ing.With the development of the high-technology,however,these dilemmas will be solved perfectly.Smart mathematics education aims to provide students with rich and modern in-quiry resources and tools through effective integration of digital,network,intelligent and multimedia information methods.As a result,students can study mathematics harder and learn mathematics better.Besides,more use of flexible education methods can make mathe-matical teaching more effective.Therefore,this thesis considers establishing a smart math-ematical education platform based on the mathematical knowledge graph to improve the mathematical education.There are two important processes in the establishment of a mathematical knowledge graph.One is mathematical entity recognition,and the other is mathematical relationship recogni-tion.As for mathematical entity recognition,it requires clarifying entities from identified entities.This thesis introduces three aspects and combines them to fuse multi-information to get bet-ter performance.Traditional mathematical entity recognition algorithms only consider one aspect in the process of recognition;this restricts the expressive power of the model to a cer-tain extent.And different mathematical entity classifications have different characteristics.So this thesis considers the effect of mathematical entity features,introducing mathematical text channel information.The definition of a mathematical entity is accurate,so it is spatially invariant.Based on the spatial invariance of mathematical entities,this thesis introduces a text space mechanism to further confirm the results.At the same time,according to the ac-curacy of the definition of mathematical entities,efficiency of Chinese word segmentation results are introduced to improve the recognition effect of word boundaries.Experiments show that the integration of information in three directions can enhance the accuracy of mathematical entity recognition to a certain extent.In this thesis,two tag ways also are in-troduced to test the influence of training data on the final performance.Experiments show that the three-digit notation way can obtain better performance.As for mathematical entity-relationship recognition,this thesis introduces two scales of ways and combines them to dig the entity-relationship.Traditional mathematical entity-relationship recognition only considers the influence of a single scale;this often prevents useful information from being fully utilized.This thesis defines eleven mathematical entity relationships,and applies them as a basis for experiments in identifying mathematical en-tity relationships.The definition of mathematical entities is logical,so this thesis considers starting from the entities themselves,so a word-scale attention mechanism is introduced to fully explore the relationships reflected by the entities themselves.At the same time,the description of mathematical problems is logical as well,so the structure of mathematical sentences also affects the relationship between entities to a certain extent.Therefore,this thesis considers introducing a sentence-scale attention mechanism to mine the relationship represented by the sentence structure.Finally,this thesis uses an adversarial model to bal-ance the performance of the two mechanisms and comprehensively obtain the final results.Experiments show that the balance of the two scales can have a significant influence on the recognition results.Based on the obtained mathematical knowledge entities and relationships,in this thesis,a mathematical knowledge graph is created and a smart education platform for students and teachers based on it is established.