HighlightCurated by Aramai EditorialarXiv (Cornell University)

Embedding Entities and Relations for Learning and Inference in Knowledge Bases

A survey on approaches to learning first-order logic rules over knowledge graphs.

This paper reviews state-of-the-art systems for learning first-order logic rules over knowledge graphs. It conducts a comparative analysis of various approaches, including ILP-based, statistical path generalisation, and neuro-symbolic methods. The authors highlight important application scenarios of rule learning in knowledge graph completion, fact checking, and other research areas.

Based on: Embedding Entities and Relations for Learning and Inference in Knowledge Bases · arXiv (Cornell University)

HighlightCurated by Aramai EditorialJournal of Artificial Intelligence Research

Text Relatedness Based on a Word Thesaurus

A paper proposing a measure of semantic relatedness between texts using word-to-word semantic links.

The authors present a new approach to measuring text relatedness based on implicit semantic links between words.,They introduce Omiotis, a measure that capitalizes on the semantic relatedness between individual words and extends it to text-to-text relatedness.,Experimental evaluation shows that this method outperforms lexicon-based methods in selected tasks.

Based on: Text Relatedness Based on a Word Thesaurus · Journal of Artificial Intelligence Research