Context-free grammar inference

Abstract

In this thesis, we study the grammatical interference task of learning context-free grammars. We restrict ourselves to the scenario where only positive data is available, which is the least amount of help that can be given to a learner. We categorise the research done in this area into four classes of techniques; two of which use the exact algorithms that are only applicable in the very restrictive scenarios and the other two use algorithms which work well on real-world data but do not guarantee any form of correctness.