A Fast Unified Model for Parsing and Sentence Understanding
This addresses slow and unwieldy parsing for large-scale NLP tasks, though it is incremental as it builds on existing tree-structured models.
The authors tackled the inefficiency of tree-structured neural networks for NLP by introducing SPINN, a unified model that integrates parsing and interpretation, achieving up to 25x speedup and outperforming other models on the Stanford NLI task.
Tree-structured neural networks exploit valuable syntactic parse information as they interpret the meanings of sentences. However, they suffer from two key technical problems that make them slow and unwieldy for large-scale NLP tasks: they usually operate on parsed sentences and they do not directly support batched computation. We address these issues by introducing the Stack-augmented Parser-Interpreter Neural Network (SPINN), which combines parsing and interpretation within a single tree-sequence hybrid model by integrating tree-structured sentence interpretation into the linear sequential structure of a shift-reduce parser. Our model supports batched computation for a speedup of up to 25 times over other tree-structured models, and its integrated parser can operate on unparsed data with little loss in accuracy. We evaluate it on the Stanford NLI entailment task and show that it significantly outperforms other sentence-encoding models.