Linguistic Frameworks Go Toe-to-Toe at Neuro-Symbolic Language Modeling

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Prange, Jakob, Schneider, Nathan, Kong, Lingpeng
Format: Preprint
Published: 2021
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918425803096064
author Prange, Jakob
Schneider, Nathan
Kong, Lingpeng
author_facet Prange, Jakob
Schneider, Nathan
Kong, Lingpeng
contents We examine the extent to which, in principle, linguistic graph representations can complement and improve neural language modeling. With an ensemble setup consisting of a pretrained Transformer and ground-truth graphs from one of 7 different formalisms, we find that, overall, semantic constituency structures are most useful to language modeling performance -- outpacing syntactic constituency structures as well as syntactic and semantic dependency structures. Further, effects vary greatly depending on part-of-speech class. In sum, our findings point to promising tendencies in neuro-symbolic language modeling and invite future research quantifying the design choices made by different formalisms.
format Preprint
id arxiv_https___arxiv_org_abs_2112_07874
institution arXiv
publishDate 2021
record_format arxiv
spellingShingle Linguistic Frameworks Go Toe-to-Toe at Neuro-Symbolic Language Modeling
Prange, Jakob
Schneider, Nathan
Kong, Lingpeng
Computation and Language
Artificial Intelligence
We examine the extent to which, in principle, linguistic graph representations can complement and improve neural language modeling. With an ensemble setup consisting of a pretrained Transformer and ground-truth graphs from one of 7 different formalisms, we find that, overall, semantic constituency structures are most useful to language modeling performance -- outpacing syntactic constituency structures as well as syntactic and semantic dependency structures. Further, effects vary greatly depending on part-of-speech class. In sum, our findings point to promising tendencies in neuro-symbolic language modeling and invite future research quantifying the design choices made by different formalisms.
title Linguistic Frameworks Go Toe-to-Toe at Neuro-Symbolic Language Modeling
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2112.07874