Semantic Role Labeling With Relative Clauses

One of the main goals of computerized linguistic studies is automatically finding the elements of a sentence. Separation of sentences with too many judgements into their elements is a more complicated process when compared with simple sentences. In this study, instead of separating the whole sentence into its elements, the separation of sub clauses into their own elements is suggested. This approach can be considered as dividing a hard problem into sub parts; and this had higher rate of achievement when compared with dividing the whole sentence into its elements. Condition Random Fields (CRF) algorithm was used for dividing the sentences automatically into subclauses and finding its elements.