Statistical Interpretation of Compound Nominalisations

This paper presents a method for detecting compound nominalisations from open data, and providing a semantic intepretation. It uses a statistical model based on confidence intervals over frequencies extracted from a large, balanced corpus. Using three paraphrases of the given compound nominalisation, and interpretation preferences of its components, the algorithm achieves about 70% accuracy in classifying the semantic relationship as one of subject, and object, and 57% between subject, direct object, and prepositional object.