Heterogeneous Formal Contexts For Uncertainty Reasoning

The study on heterogeneous data of information resources has been the focus in thedatabase field. A lot of valuable information in heterogeneous data need to be mined anddeveloped. This thesis studies fuzzy formal context which includes heterogeneous values.Since conversion and unification for heterogeneous data are the emphasis and di?culty inthe study of heterogeneous formal contexts, three di?erent methods are used to processthe heterogeneous data. We mainly study attribute classification, multi-adjoint reasoningand information fusion in heterogeneous formal contexts.The main work of this thesis can be summarized as follows:1. Heterogeneous attribute set is indicated by fuzzy attribute subsets in heteroge-neous formal contexts. In this case, the heterogeneous information can be transformedand expressed by fuzzy attribute subsets, and the problem of attributes heterogeneous canbe processed properly. Then the attribute classification in the rough set theory is extend-ed to heterogeneous formal contexts, which is managed through meet-irreducible element.Thus the characteristics of three types of attribute classification in heterogeneous formalcontexts are obtained.2. The multi-adjoint L-connected theory is applied to heterogeneous formal con-texts. Heterogeneous data of objects and attributes in heterogeneous formal contextsare evaluated on the same lattice L by L-connected, which is through a specific mapping.Then a new definition of heterogeneous L-connected concept is introduced and a completeheterogeneous L-connected concept lattice is constructed.3. The idea of evidence theory is introduced to heterogeneous formal contexts. Onthe premise of maintaining the independent data, the evaluation measure is defined foreach pair of objects and heterogeneous attributes in order to realize the measurementunification of heterogeneous data. Then the combination formula is applied to composethe information of heterogeneous formal contexts, and the combination algorithm of theevaluation measure is formulated. Finally an example is employed to illustrated our fusionmethod in the heterogeneous formal contexts.