Discovering Specific Semantic Relationships between Nouns and Verbs in a Specialized French Corpus
Vincent Claveau, Marie-Claude L’Homme · 2004
Recent literature in computational terminology has shown an increasing interest in identifying various semantic relationships between terms. In this paper, we propose an original strategy to find specific noun-verb combinations in a specialized corpus. We focus on verbs that convey a meaning of realization. To acquire these noun-verb pairs, we use ASARES, a machine learning technique that automatically infers extraction patterns from examples and counter-examples of realization noun-verb pairs. The patterns are then applied to the corpus to retrieve new pairs. Results, measured with a large test set, show that our acquisition technique outperforms classical statistical methods used for collocation acquisition. Moreover, the inferred patterns yield interesting clues on which structures are more likely to convey the target semantic link.