The Methods Of Association Rule Extraction Based On Linguistic-valued Formal Context

Uncertain and fuzzy information exists in human perception of the real world,which is usually described in linguistic values.The linguistic values can express qualitative knowledge,which is closer to the expression of people's daily life.It can better describe the fuzziness and uncertainty of human thinking.Formal concept analysis is a powerful tool for data analysis and rule extraction.It has great research value,using the theory of formal concept analysis to deal with linguistic-valued information system and study the method of association rule extraction based on linguistic-valued formal context.Based on the linguistic-valued formal context,the conversion method of linguistic-valued formal context with different granularity and association rules extraction methods are studied in this thesis.The main research contents include:1.Aiming at the problem that different people usually use different granularity linguistic term sets when describing the same thing,the conversion method of linguistic-valued formal context with different granularity is studied.By defining and maintaining the normalized distance between the linguistic terms,the conversion method of linguistic-valued formal context with different granularity is proposed.The process of the conversion method is reversible and can avoid the loss of information.Dynamic conversion of linguistic-valued formal context under different constraints is realized by adjusted the threshold.2.The traditional method of extracting association rules gets a lot of redundant rules and cannot show the frequent relationship between data directly.Therefore,an association rule extraction method based on the fuzzy linguistic attribute ordered structure diagram is proposed.The fuzzy linguistic attribute ordered structure diagram is constructed by introducing fuzzy linguistic values into the building process.Mining the maximum frequent linguistic item sets at the same support and extracting the non-redundant association rules from the fuzzy linguistic attribute ordered structure diagram.The association rules are transformed into the knowledge representation in the fuzzy linguistic attribute ordered structure diagram,which is expressed by association nodes and association paths,and the visualization of association rules is realized.An example of student evaluation is given to illustrate the effectiveness of the proposed method.3.In response to the incomplete linguistic concept context,an uncertain information completion method is proposed in this thesis.A method of extracting association rules based on quantitative linguistic concept is proposed by defining quantitative linguistic concept lattice.For the incomplete linguistic concept formal context,the uncertain concepts are obtained by introducing uncertain information to missing information.People can set the threshold of uncertain information according to the requirements of different situations.Different concepts can be obtained under different thresholds to extract different results of association rules.