CMILLS: Adapting Semantic Role Labeling Features to Dependency Parsing

Chad Mills,Gina-Anne Levow

doi:10.18653/v1/s15-2075

Abstract

We describe a system for semantic role labeling adapted to a dependency parsing framework. Verb arguments are predicted over nodes in a dependency parse tree instead of nodes in a phrase-structure parse tree. Our system participated in SemEval-2015 shared Task 15, Subtask 1: CPA parsing and achieved an Fscore of 0.516. We adapted features from prior semantic role labeling work to the dependency parsing paradigm, using a series of supervised classifiers to identify arguments of a verb and then assigning syntactic and semantic labels. We found that careful feature selection had a major impact on system performance. However, sparse training data still led rule-based systems like the baseline to be more effective than learning-based approaches.

Highlights

We describe our submission to the SemEval-2015 Task 15, Subtask 1 on Corpus Pattern Analysis (Baisa et al 2015). This task is similar to semantic role labeling but with arguments based on nodes in dependency parses instead of a syntactic parse tree
Consider the sentence “But he said Labour did not agree that Britain could or should abandon development, either for itself or for the developing world.”
This was different from previous Semantic Role Labeling (SRL) tasks where a node in the parse tree was taken as the argument; this is more similar to identifying the headword of the phrase that’s an argument rather than identifying the full phrase

Summary

Introduction

We describe our submission to the SemEval-2015 Task 15, Subtask 1 on Corpus Pattern Analysis (Baisa et al 2015). This task is similar to semantic role labeling but with arguments based on nodes in dependency parses instead of a syntactic parse tree. Consider the sentence “But he said Labour did not agree that Britain could or should abandon development, either for itself or for the developing world.”. This subtask involves taking that sentence and making the following determinations relative to the given verb “abandon”: “Britain” is the syntactic subject of “abandon”. Falls under the “Institution” semantic type “development” is the syntactic object of “abandon” and is of semantic type “Activity” We organize the remainder of our paper as follows: Section 2 describes our system, Section 3 presents experiments, and Section 4 concludes

System Description

Featurization

Classifiers

Semantics Consistency Heuristic Filter

Experiments

Findings

Conclusion