Abstract : We investigate learning dependency grammar from partial data and membership queries as a model of natural language acquisition. We define a learning paradigm based on a dialogue between the learner and a referent who knows the target language. This dialogue consists in a presentation of structured by the learner. We define an efficient algorithm corresponding to this paradigm and illustrate it on examples.
Jérôme Besombes, Jean-Yves Marion. Learning Dependency Languages from a Teacher. 9th conference on Formal Grammar, 2004, Nancy, France, 7 p. ⟨inria-00101085⟩