On Tree-Based Neural Sentence Modeling
This challenges the assumption that syntactic guidance is crucial in NLP models, potentially simplifying encoder design for researchers and practitioners.
The study investigated whether syntactic parsing trees are essential for tree-based neural sentence encoders by replacing them with trivial trees lacking syntax, finding that trivial trees achieved competitive or better results across ten downstream tasks, indicating that explicit syntax may not be the main factor in performance gains.
Neural networks with tree-based sentence encoders have shown better results on many downstream tasks. Most of existing tree-based encoders adopt syntactic parsing trees as the explicit structure prior. To study the effectiveness of different tree structures, we replace the parsing trees with trivial trees (i.e., binary balanced tree, left-branching tree and right-branching tree) in the encoders. Though trivial trees contain no syntactic information, those encoders get competitive or even better results on all of the ten downstream tasks we investigated. This surprising result indicates that explicit syntax guidance may not be the main contributor to the superior performances of tree-based neural sentence modeling. Further analysis show that tree modeling gives better results when crucial words are closer to the final representation. Additional experiments give more clues on how to design an effective tree-based encoder. Our code is open-source and available at https://github.com/ExplorerFreda/TreeEnc.