Query Based Learning of XPath Fragments

XPath is a language for XML document navigation. It is part of many XML related applications and standards such as XSLT, Document Object Models, XQuery, XLink, etc. This paper studies active forms of learning XPath expressions. In particular, we show nonlearnability of the XPath fragments using the equivalence and membership queries. In response to these negative results and visualisation problems we de ne more suitable tree-pre x queries and present a polynomial learning algorithm which uses them.