The Cognitive Interactionist Approach of Sentence Parsing with Simple Recurrent Networks

Abstract

Sentence parsing has a long research history in the fields of machine learning and natural language processing. The state-of-the-art techniques for tackling this task are mostly based on statistical language learning. How human parsing sentences is also an important research topic attracting research efforts for decades in the field of cognitive psychology. Some behavioristic experiments have convinced that the interactionist approach is rational and effective to simulate human parsing mechanism. This paper presents a sentence parser, the Interactionist Parser, which incorporated the cognitive interactionist approach with semantic information and simple recurrent networks, to extend and enrich the techniques for sentence parsing. Thinking of the parsing efficiency, the semantic information of two word types, noun and verb, are included during the parsing procedure in current stage. The experimental results demonstrate that the Interactionist Parser has comparability with the state-of-the-art parsing techniques based on statistical language learning.