Incorporating Linguistics Constraints into Inductive Logic Programming

We report work on effectively incorporating lin-guistic knowledge into grammar induction. Weuse a highly interactive bottom-up inductivelogic programming (ILP) algorithm to learn'missing' grammar rules from an :incompletegrammar. Using linguistic constraints on, forexample, head features and gap threading, re-duces the search space to such an extent that,in the small-scale experiments reported here,we can generate and store all candidate gram-mar rules together with information about theircoverage and linguistic properties. This allowsan appealingly simple and controlled methodfor generating linguistically plausible grammarrules. Starting from a base of highly spe-cific rules, we apply least general generalisationand inverse resolution to generate more generalrules. Induced rules are ordered, for example bycoverage, for easy inspection by the user and atany point, the user can commit to a hypothe-sised rule and add it to the grammar. Relatedwork in ILP and computational linguistics isdiscusse