Towards Inference of More Realistic XSDs

The XML has undoubtedly become a standard for data rep-resentation and manipulation. But most of XML documents are still created without the respective description of their structure, i.e. an XML schema. Hence, in this paper we fo-cus on the problem of automatic inferring of an XML schema for a given sample set of XML documents. Contrary to ex-isting works, whose aim is to infer as concise schema as pos-sible, we focus on inferring of a more realistic result, i.e. a schema that is closer to human-written ones and bears more precise information. For this purpose we extend and com-bine the existing verified techniques (such as ACO heuristics or MDL principle) with a set of heuristics exploiting seman-tics of element/attribute names, thesauri or statistical anal-ysis of input data. Using a set of examples we show and discuss advantages of our proposal