With the rise of hyperlinked networks and the broad spectrum of information that is accessible over the Web, a lot of data sources become available. XML standard receives large acceptance, the comparison of the Web to a "database" is closer to reality than ever before. One of the long-standing and challenging problems for many applications, is schema matching. While there are many schema-mathing algorithms propsed in the literature to tackle this problem, most of them are not tackling the indirect mappings, just the few ones do, but still there is a need to cover the spesial cases of XML schemas. In this paper, we propose a new composite approach that detects direct and indirect mappings for XML schemas with higher accuracy, by proposing agorithms for XML schema matching, after combing them, to detect most kinds of mapping. Our approach to automating with the data value characteristics and the expected data values for the element-level matching, which combined with the XML structural mappings by appliying a dedicated XML Match taxonomy. We also,develop an ontology that fits to the structure of XML schemas, in order to solve semantic conflicts.