Application Of Data Mining In Ancient Ceramic Dating And Ceramic Raw Material Classification

Currently,with the rapid development of data mining and the vigorous rise of the ceramic industry,the application research of data mining in the ceramic industry has increasingly become a hot topic in the current industry.The application research of data mining in ancient ceramic dating and ceramic raw material classification has also focused on the field of data mining and the vision of experts and scholars in the ceramic industry.In the past,in the dating of ancient ceramics,based on expert identification experience,the dating of ancient ceramics was determined through hand touch,visual inspection,and other sensory functions.However,due to personal subjective factors,it was inevitable that the identification results were "expert oriented";In ceramic raw material classification,ceramic raw materials are classified based on past experience in ceramic raw material classification.Although there can be a general classification of ceramic raw materials,the classification is not accurate,precise,and reasonable.In the application research of data mining to ancient ceramic dating and ceramic raw material classification,this topic is committed to improving the above existing problems.At the same time,it has also conducted some exploration of new models that can be applied in this field.It has made some small attempts to find data mining algorithms that are more suitable for ancient ceramic dating and ceramic raw material classification.Based on previous research,it provides a little experience for future researchers.Aiming at the problems existing in the past methods of ancient ceramic dating and ceramic raw material classification,this paper proposes five data mining methods based on the chemical composition of ancient ceramics and ceramic raw materials to conduct application research on ancient ceramic dating and ceramic raw material classification.In the application research of data mining to the dating of ancient ceramics,the chemical composition content data of 17 types of ancient ceramics obtained from the ceramic research institute were used as predictive variables,and the ancient ceramic categories of the Yuan,Ming,and Qing dynasties were used as target outputs.Two algorithms,MLP neural network and RBF neural network,were used to establish models for the dating analysis of ancient ceramics,and the dating accuracy rates obtained were 76.79% and85.71%,respectively;In the application research of data mining to the classification of ceramic raw materials,the chemical composition data of ceramic raw materials collected from "Ceramic Technology" are used as predictive variables to input,and three different categories of ceramic raw materials are used as target outputs.The random forest algorithm,Markov covering structure method based on Bayesian network Based on the tree expansion naive Bayesian algorithm,ceramic raw material classification models were established for ceramic raw material classification experiments,and the classification accuracy rates were 78.57%,85.71%,and 92.86%,respectively.Based on the above experimental results,it is concluded that in the dating analysis of ancient ceramics,the accuracy of RBF neural network is higher than that of MLP neural network;In the classification experiment of ceramic raw materials,the classification prediction effect of tree extended naive Bayesian algorithm is better than that of Markov cover structure method based on Bayesian network and random forest algorithm.