Transductive Parsing for Universal Decompositional Semantics

10/22/2019
by   Elias Stengel-Eskin, et al.
0

We introduce a transductive model for parsing into Universal Decompositional Semantics (UDS) representations, which jointly learns to map natural language utterances into UDS graph structures and annotate the graph with decompositional semantic attribute scores. We also introduce a strong pipeline model for parsing UDS graph structure, and show that our parser can perform comparably while additionally performing attribute prediction.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
04/27/2017

Learning Structured Natural Language Representations for Semantic Parsing

We introduce a neural semantic parser that converts natural language utt...
research
09/10/2019

Core Semantic First: A Top-down Approach for AMR Parsing

We introduce a novel scheme for parsing a piece of text into its Abstrac...
research
03/14/2020

Universal Function Approximation on Graphs using Multivalued Functions

In this work we produce a framework for constructing universal function ...
research
05/24/2022

GraphQ IR: Unifying Semantic Parsing of Graph Query Language with Intermediate Representation

Subject to the semantic gap lying between natural and formal language, n...
research
10/16/2013

ImageSpirit: Verbal Guided Image Parsing

Humans describe images in terms of nouns and adjectives while algorithms...
research
04/09/2019

Graphonomy: Universal Human Parsing via Graph Transfer Learning

Prior highly-tuned human parsing models tend to fit towards each dataset...
research
11/09/2020

A Semantic Framework for PEGs

Parsing Expression Grammars (PEGs) are a recognition-based formalism whi...

Please sign up or login with your details

Forgot password? Click here to reset