A Step toward Machine Recognition of Complex Sentences

This paper presents theoretical and technological background of a model for machine recognition of complex sentences. It is based on the Syntactic and Semantic Framework (SSF) which implements fundamental linguistic fields network resources and encyclopedias. It can be used to extract subject, predicate and object, as well as other sentence's parts (e.g. NP/VP/PP), and in some cases even semantic roles. In compound sentences the machine can easily recognize independent sentences, whereas in complex sentences the machine recognizes the main clause and the related subordinate clauses as well as sentence types (subject, object, predicate, etc.). Using stored patterns various theories can be tested.