Three New Probabilistic Models for Dependency Parsing: An Exploration

TL;DR

This paper introduces a new cubic-time dependency parsing algorithm and explores three probabilistic models, with the generative model showing superior performance on Wall Street Journal data.

Contribution

It presents a novel O(n^3) parsing algorithm and three contrasting stochastic models for dependency parsing, highlighting the effectiveness of the generative approach.

Findings

The generative model outperforms the others in parsing accuracy.

All models perform similarly in part-of-speech tagging.

Preliminary results are based on Wall Street Journal data.

Abstract

After presenting a novel O(n^3) parsing algorithm for dependency grammar, we develop three contrasting ways to stochasticize it. We propose (a) a lexical affinity model where words struggle to modify each other, (b) a sense tagging model where words fluctuate randomly in their selectional preferences, and (c) a generative model where the speaker fleshes out each word's syntactic and conceptual structure without regard to the implications for the hearer. We also give preliminary empirical results from evaluating the three models' parsing performance on annotated Wall Street Journal training text (derived from the Penn Treebank). In these results, the generative (i.e., top-down) model performs significantly better than the others, and does about equally well at assigning part-of-speech tags.