Fast-and-Frugal Text-Graph Transformers are Effective Link Predictors

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Coman, Andrei C., Theodoropoulos, Christos, Moens, Marie-Francine, Henderson, James
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916794583744512
author Coman, Andrei C.
Theodoropoulos, Christos
Moens, Marie-Francine
Henderson, James
author_facet Coman, Andrei C.
Theodoropoulos, Christos
Moens, Marie-Francine
Henderson, James
contents We propose Fast-and-Frugal Text-Graph (FnF-TG) Transformers, a Transformer-based framework that unifies textual and structural information for inductive link prediction in text-attributed knowledge graphs. We demonstrate that, by effectively encoding ego-graphs (1-hop neighbourhoods), we can reduce the reliance on resource-intensive textual encoders. This makes the model both fast at training and inference time, as well as frugal in terms of cost. We perform a comprehensive evaluation on three popular datasets and show that FnF-TG can achieve superior performance compared to previous state-of-the-art methods. We also extend inductive learning to a fully inductive setting, where relations don't rely on transductive (fixed) representations, as in previous work, but are a function of their textual description. Additionally, we introduce new variants of existing datasets, specifically designed to test the performance of models on unseen relations at inference time, thus offering a new test-bench for fully inductive link prediction.
format Preprint
id arxiv_https___arxiv_org_abs_2408_06778
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Fast-and-Frugal Text-Graph Transformers are Effective Link Predictors
Coman, Andrei C.
Theodoropoulos, Christos
Moens, Marie-Francine
Henderson, James
Computation and Language
We propose Fast-and-Frugal Text-Graph (FnF-TG) Transformers, a Transformer-based framework that unifies textual and structural information for inductive link prediction in text-attributed knowledge graphs. We demonstrate that, by effectively encoding ego-graphs (1-hop neighbourhoods), we can reduce the reliance on resource-intensive textual encoders. This makes the model both fast at training and inference time, as well as frugal in terms of cost. We perform a comprehensive evaluation on three popular datasets and show that FnF-TG can achieve superior performance compared to previous state-of-the-art methods. We also extend inductive learning to a fully inductive setting, where relations don't rely on transductive (fixed) representations, as in previous work, but are a function of their textual description. Additionally, we introduce new variants of existing datasets, specifically designed to test the performance of models on unseen relations at inference time, thus offering a new test-bench for fully inductive link prediction.
title Fast-and-Frugal Text-Graph Transformers are Effective Link Predictors
topic Computation and Language
url https://arxiv.org/abs/2408.06778