The Research And Application Of Machine Learning Algorithms Based On Triadic Concept Analysis

Triadic concept analysis(TCA)is a novel method of data analysis.It is an effective extension of three-dimensional data or multi-dimensional data processing based on the theory of formal concept analysis(FCA).It is also adapting to the application of diversity data in the era of big data.At present,there are relatively few studies on TCA,and the analysis of the triadic concept to construct the learning model framework is also relatively scarce.Therefore,how to use the triadic context to effectively extract the triadic concept and how to make the model can be applied,etc,which have become a primary problem which decides the concepts could be successfully applied.In this research work,the MLTCA(Machine Learning Triadic Concept Analysis)model is proposed and validated.The specific work are as follows:First,the property of triadic concept is analyzed by the basic theory of the triadic concept analysis and the Tri-concepts algorithm is proposed to construct the triadic concept.The algorithm is based on the triadic context to extract the concept of the tri-generator from the three sets of objects,attributes and conditions.After the closure operator of the h induced operator,the tri-sets are gradually generated.Then all the conditions of the tri-sets are merged and the redundant tri-sets are deleted to get the classical triadic concept eventually.Then based on the Tri-concepts algorithm to consider the TCA for the practical application of the requirements.In order to make use of the triadic concept to express the data of the real application,it is proposed to combine the TCA with the fuzzy set and define the triadic concept with the membership degree,that is,the binary relation in the context is extended to a fuzzy relation of the interval which represents membership degree about attribute for object under certain conditions.So the proposed model can construct the concept with membership degree from the original information.According to the representation of the close degree in the fuzzy theory,the similarity between the triadic concepts is obtained,and then the similarity between the concept of the training data and the test data can be calculated,thusreaching the purpose of classification.Hence,this research proposes a learning model from the data representation to the reasoning classification.Finally,the model is applied to the text classification to illustrate its correctness and validity.Experiments on different data sets show that the model has good performance on a particular data and can be efficiently classified.