Entity Type Prediction Leveraging Graph Walks and Entity Descriptions

TL;DR

This paper introduces GRAND, a new method for entity typing in knowledge graphs that combines graph walk strategies with textual descriptions, significantly improving accuracy over existing methods.

Contribution

GRAND innovatively integrates graph walk strategies with textual entity descriptions using RDF2vec and language models for enhanced entity typing.

Findings

Outperforms baseline methods on DBpedia and FIGER datasets.

Order-aware RDF2vec variants improve embedding quality.

Combining graph walks with textual descriptions yields best results.

Abstract

The entity type information in Knowledge Graphs (KGs) such as DBpedia, Freebase, etc. is often incomplete due to automated generation or human curation. Entity typing is the task of assigning or inferring the semantic type of an entity in a KG. This paper presents \textit{GRAND}, a novel approach for entity typing leveraging different graph walk strategies in RDF2vec together with textual entity descriptions. RDF2vec first generates graph walks and then uses a language model to obtain embeddings for each node in the graph. This study shows that the walk generation strategy and the embedding model have a significant effect on the performance of the entity typing task. The proposed approach outperforms the baseline approaches on the benchmark datasets DBpedia and FIGER for entity typing in KGs for both fine-grained and coarse-grained classes. The results show that the combination of order-aware RDF2vec variants together with the contextual embeddings of the textual entity descriptions achieve the best results.