A version space algorithm for learning extracting rules from tree-structured documents. Resulting hypotheses are called "tag tree patterns". A Mapping from tag tree patterns to statements in the XPath language is defined.
M. Keijzer, M. Baptist, V. Babovic, and J. Uthurburu. GECCO 2005: Proceedings of the 2005 conference on
Genetic and evolutionary computation, 2, page 1999--2006. Washington DC, USA, ACM Press, (25-29 June 2005)