Cooperative Learning of Disjoint Syntax and Semantics
This addresses a fundamental issue in computational linguistics for researchers and practitioners by improving parsing accuracy, though it appears incremental as it builds on prior recursive models.
The paper tackles the problem of models failing to learn correct parsing strategies for mathematical expressions from a simple context-free grammar, achieving near perfect accuracy on this task.
There has been considerable attention devoted to models that learn to jointly infer an expression's syntactic structure and its semantics. Yet, \citet{NangiaB18} has recently shown that the current best systems fail to learn the correct parsing strategy on mathematical expressions generated from a simple context-free grammar. In this work, we present a recursive model inspired by \newcite{ChoiYL18} that reaches near perfect accuracy on this task. Our model is composed of two separated modules for syntax and semantics. They are cooperatively trained with standard continuous and discrete optimization schemes. Our model does not require any linguistic structure for supervision and its recursive nature allows for out-of-domain generalization with little loss in performance. Additionally, our approach performs competitively on several natural language tasks, such as Natural Language Inference or Sentiment Analysis.