A comparative study on schema matching membership functions using a neuro-fuzzy approach

摘要

The semantic correspondence among schemas can be found using the toolset provided by schema matching. The classic schema matching is based on crisp set theory, but ongoing researches are targeting the fuzzy schema matching. In a previous study, we proposed schema matching threshold functions, which were later shown to be related to the concept of fuzzy membership functions. In this paper, we advance one step further in this field and assess the goodness of various types and numbers of membership functions for a schema matcher.