Extracting Novel Facts from Tables for Knowledge Graph Completion (Extended version)

TL;DR

This paper introduces a novel end-to-end approach for extracting more unique facts from tables to enhance knowledge graph completion, outperforming existing methods in recall and novelty.

Contribution

It presents a scalable graphical model with entity similarity and KG embeddings for disambiguation, enabling more novel fact extraction without assumptions about the KG.

Findings

Higher recall in table interpretation compared to state-of-the-art

Produces more novel facts, reducing redundancy

More resistant to bias in fact extraction

Abstract

We propose a new end-to-end method for extending a Knowledge Graph (KG) from tables. Existing techniques tend to interpret tables by focusing on information that is already in the KG, and therefore tend to extract many redundant facts. Our method aims to find more novel facts. We introduce a new technique for table interpretation based on a scalable graphical model using entity similarities. Our method further disambiguates cell values using KG embeddings as additional ranking method. Other distinctive features are the lack of assumptions about the underlying KG and the enabling of a fine-grained tuning of the precision/recall trade-off of extracted facts. Our experiments show that our approach has a higher recall during the interpretation process than the state-of-the-art, and is more resistant against the bias observed in extracting mostly redundant facts since it produces more novel extractions.