Research On Concept Lattice Updating Construction Algorithms Based On Concept Hierarchy

Concept lattice is an effective tool for knowledge discovering and data mining. The process of forming concept lattice from formal context is actually a concept clustering process. Concept lattice reflects the relationship of generalization and specialization among concepts. How to update concept lattice effectively and improve the construction efficiency have great significance while the concept hierarchy of attributes change in the formal context. An effective updating algorithm is presented according to the changes of attributes concept hierarchy to improve the updating construction efficiency of concept lattice.The main researches are as follows:(1) An UCP algorithm of concept lattice updating construction based on concept promoting is proposed. When multi-attribute in the formal context is combining (concept promotion or generalization), some concepts in concept lattice are promoted or specialized from a low level to a higher one. By making use of the concept lattice constructed before concept promotion, the concept only containing combination attributes in the intent is operated. If the intersection of concept intent and combination attribute is null, then do not operate anything, else, newly added concept or update concept is built according to the different intersection. Then, redundancy concepts which appearing during update process are combined or deleted and the concept lattice of new formal context is generated. Finally, by taking discrete celestial spectrum data as the formal context and comparing with the Godin algorithm, the experiment results show the correctness and the validity of the UCP algorithm.(2)An UCCS algorithm of concept lattice updating construction based on concept specialization is proposed. This algorithm mainly aims at problems which single attribute is divided into multi-attribute in the formal context, namely some concepts in the concept lattice are specialized or exemplified from higher level ones to lower level ones. Firstly, by making use of new attributes divided by attribute, concept lattice is constructed corresponding. Then, the concept lattice is compared with a sub-concept lattice of former concept lattice, and the updated concept, newly added concept and specialization concept are generated. Because of not comparing with the former concept lattice, the time is saved and the updating construction efficiency of concept lattice is improved. Finally, experiment results show the correctness and the validity of the UCCS algorithm by taking the discrete celestial spectrum data as the formal context.(3)The updating construction prototype system of the concept lattice is designed based on concept levels. In the Windows XP operating system, by taking VC++6.0 and ORACLE9i as development tools, using discrete celestial spectrum data as formal context, design and realize concept lattice updating construction prototype system based on concept hierarchy .