Research On Key Technologies In Formal Concept Analysis With Constrained Relationships

The development of big data is still the focus of the 14 th Five-Year Plan.Big data is not only an emerging technology industry,but also an element,resource and power to integrate into various fields of economic and social development.Data mining is an important part in the process of big data application.How to extract the information that people are interested in from a large number of complex data has become a research hotspot.As a powerful tool for data analysis and knowledge representation,formal concept analysis has been successfully applied in data mining.The data structure of formal context,which consists of objects,attributes and binary relationships between objects and attributes,is the basis of theoretical research and application of formal concept analysis,and its role is irreplaceable.At present,research of formal concept analysis usually generates formal context from domain knowledge,and induces corresponding concepts and concept lattice from formal context,and applies them in various fields.However,in the process of generating formal context,there are incomplete objects due to different understanding of data or limited acquisition methods.In addition,the dynamic changes of data,formal context and concept lattice need to be updated simultaneously.Researchers pay more attention to concept lattice updating of one object or attribute changes,and less attention to dynamic updating of batch objects and attributes.Based on the above considerations,this paper focuses on the theory and application of formal concept analysis with constrained relationships.（1）The relationship between attributes is analyzed and discussed,and the formal context generation algorithm is given.This paper studies the relationship between attribute prerequisite and attribute implication,and proves that attribute prerequisite is a special attribute implication.Based on the idea of backward elimination,a reduction algorithm of formal context generation based on attribute implication is proposed.Based on the idea of forward regression,this paper proposes an expansion algorithm of formal context generation based on attribute implication.By finding a set of object bases with a given attribute implication set,the algorithm gradually expands to generate formal context.The effectiveness of the algorithm is verified by theoretical proof and example analysis.（2）The change rule of nodes of concept lattice before and after object set deleting is analyzed and discussed,and the updating algorithm of concept lattice for batch object and attribute deleting is given.Top-down of object and bottom-up of attribute are used to update concept lattice synchronously.Theoretical analysis and example analysis verify the effectiveness of the algorithm.（3）An ontology construction model based on formal concept analysis of attribute implication is proposed and applied to the field of film resources.This paper introduces the construction method of the model in detail,and takes douban web page as an example to realize the construction and update of the domain ontology of film resources.