Saved in:
Bibliographic Details
Main Authors: Zeng, Peter, Stortz, Hannah, Sclafani, Eric, Shabaeva, Alina, Garza, Maria Elizabeth, Greeson, Daniel, Rambow, Owen
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2406.12131
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914171003600896
author Zeng, Peter
Stortz, Hannah
Sclafani, Eric
Shabaeva, Alina
Garza, Maria Elizabeth
Greeson, Daniel
Rambow, Owen
author_facet Zeng, Peter
Stortz, Hannah
Sclafani, Eric
Shabaeva, Alina
Garza, Maria Elizabeth
Greeson, Daniel
Rambow, Owen
contents We present Gram2Vec, a grammatical style embedding system that embeds documents into a higher dimensional space by extracting the normalized relative frequencies of grammatical features present in the text. Compared to neural approaches, Gram2Vec offers inherent interpretability based on how the feature vectors are generated. In this paper, we use authorship verification and AI detection as two applications to show how Gram2Vec can be used. For authorship verification, we use the features from Gram2Vec to explain why a pair of documents is by the same or by different authors. We also demonstrate how Gram2Vec features can be used to train a classifier for AI detection, outperforming machine learning models trained on a comparable set of Biber features.
format Preprint
id arxiv_https___arxiv_org_abs_2406_12131
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Gram2Vec: An Interpretable Document Vectorizer
Zeng, Peter
Stortz, Hannah
Sclafani, Eric
Shabaeva, Alina
Garza, Maria Elizabeth
Greeson, Daniel
Rambow, Owen
Computation and Language
We present Gram2Vec, a grammatical style embedding system that embeds documents into a higher dimensional space by extracting the normalized relative frequencies of grammatical features present in the text. Compared to neural approaches, Gram2Vec offers inherent interpretability based on how the feature vectors are generated. In this paper, we use authorship verification and AI detection as two applications to show how Gram2Vec can be used. For authorship verification, we use the features from Gram2Vec to explain why a pair of documents is by the same or by different authors. We also demonstrate how Gram2Vec features can be used to train a classifier for AI detection, outperforming machine learning models trained on a comparable set of Biber features.
title Gram2Vec: An Interpretable Document Vectorizer
topic Computation and Language
url https://arxiv.org/abs/2406.12131