Extracting Information from Participial Structures

Our applied linguistic research aims at increasing the efficiency of a rule-based information extraction (IE) system by enhancing it with further grammatical knowledge. The input of the IE system is made up of sentences of business news. The event of the piece of news is identified through the main verb of the sentence, while participants and circumstances of the event are identified through arguments and adjuncts of the main verb. Our objective was to unfold the hidden information, contained by NP's within which non-finite verbs (e.g. participles) appear. Thus, we invented a rule-system to transform participial structures into sentences with a finite verb, so that they could serve as input to the IE system. To tackle this task we had to be able to distinguish between real participles and adjectives. According to us there are some distributional criteria which can be used as the basis for creating the right classification. Adapted from the source document