A Hybrid Method to Determine Inter-Clause Dependency

Abstract

The longer the input sentences, the worse the syntactic parsing results. Therefore, a long sentence is first divided into several clauses, and the syntactic analysis for each clause is performed. Finally, all the analysis results are merged into one. In the merging process, it is difficult to determine the dependency among clauses. To handle such syntactic ambiguity in determining inter-clause dependency, this paper proposes a hybrid method using restriction rules and Decision Trees-based machine learning. Based on restriction rules, clauses that cannot be the governor of a dependent clause are excluded from the governor candidates. Next, using Decision Trees machine learning algorithm, one clause is selected as the governor of a dependent clause. We extract various features from a clause, and analyze the effect of each feature on the performance. Experimental results show that our hybrid method outperformed both methods of using restriction rules and using Decision Trees.