SBP: Preparation of Schema Elements to Schema Matching

Schema pre-matching is very critical for having schema elements fully ready for schema matching. Ideally, words are first extracted from the schema elements’ labels, and then the semantically corresponding elements are generated accordingly. Searching for the sense of words based on their vertical and horizontal contexts, and before performing schema matching is very crucial to obtain high matching accuracy and as a result increase the amount of accurate matches and reduce the number of inaccurate matches and missed matches. Nonetheless, this problem is much more challenging than it seems. This is because complete and precise information about the meaning behind each element is often unavailable. This paper presents SBP, a Sense-Based Pre-matching approach designed for hierarchical data representations. SBP consists of two main components. First, the words sets generator generates, from each element, a set of words that fully describes its meaning. Second, the words qualifier identifies the senses of words based on both their vertical and horizontal contexts. Experimental results on real-world domains show high matching accuracy obtained when using the SBP approach. © 2021, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.