Generating collaborative rule bases using fuzzy c-means with feature partitions

This paper reviews several recent papers on dataset collaborations and examines propitious approaches to the design of collaborative rule bases. The generation of collaborative rule bases using fuzzy c-means with feature partitions (FCMP) is discussed in particular. When optimizing dataset integration for a rule base it is important to identify the mode of collaboration between datasets. Unique samples and/or features in distinct datasets suggest non-uniform contributions from these datasets to the final rule base. Other considerations include sample accuracy, preserving privacy and retaining industry advantage (collaborators may be inclined to employ abstraction mechanisms on the datasets before pooling the data). These considerations demand that datasets be associated with a quality rating. A simple example using census data demonstrates the generation of collaborative rules.