Template Kernels for Dependency Parsing
TL
TL;DR
This work presents a novel kernel which facilitates efficient parsing with feature representations corresponding to a much larger set of combinations, and integrates into a parse reranking system and demonstrates its effectiveness on four languages from the CoNLL-X shared task.
Abstract
A common approach to dependency parsing is scoring a parse via a linear function of a set of indicator features. These features are typically manually constructed from templates that are applied to parts of the parse tree. The templates define which properties of a part should combine to create features. Existing approaches consider only a small subset of the possible combinations, due to statistical and computational efficiency considerations. In this work we present a novel kernel which facilitates efficient parsing with feature representations corresponding to a much larger set of combinations. We integrate the kernel into a parse reranking system and demonstrate its effectiveness on four languages from the CoNLL-X shared task. 1
References24 items
1
How to Scale Up Kernel Methods to Be As Good As Deep Neural Nets
2
Greed is Good if Randomized: New Inference for Dependency Parsing
3
Low-Rank Tensors for Scoring Dependency Structures
4
A Feature-Enriched Tree Kernel for Relation Extraction
5