Vecsigrafo: Corpus-based Word-Concept Embeddings - Bridging the Statistic/Symbolic Representational Gap

Tracking #: 1864-3077

This paper is currently under review
José Manuel Gómez-Pérez
Ronald Denaux1

Responsible editor: 
Guest Editors Semantic Deep Learning 2018

Submission type: 
Full Paper
The proliferation of knowledge graphs and recent advances in Artificial Intelligence have raised great expectations related to the combination of symbolic and distributional semantics in cognitive tasks. This is particularly the case of knowledge-based approaches to natural language processing as near-human symbolic understanding and explanation rely on expressive structured knowledge representations that tend to be labor-intensive, brittle and biased. This paper reports research addressing such limitations by capturing as embeddings in a joint space both words and concepts from large document corpora. We compare the quality of the resulting embeddings and show that they outperform word-only embeddings for a given corpus.
Full PDF Version: 
Under Review