Title: Knowledge Graphs for Natural Language Processing

Speaker: José Manuel Gómez-Pérez

Abstract

Knowledge graphs provide structured representations of entities, concepts and relations that are useful in natural language processing for tasks like word-sense disambiguation, natural language inference, entailment or classification. In this class we will show how knowledge graphs can contribute to train more expressive models that are able to learn the meaning of words by linking them to explicitly represented concepts in the graph. We will illustrate how this can be achieved by jointly learning word and concept representations from both the knowledge graph and a text corpus as embeddings in a common vector space. In addition to increased accuracies for tasks in natural language processing we will see how this approach also entails benefits for knowledge graph curation and evolution, like suggesting new concepts in a knowledge graph, merging previously existing concepts, and supporting the alignment of existing knowledge graphs.

The slides for the presentation are available here.

The author has generously arranged for a discount coupon for his book titled A Practical Guide to Hybrid Natural Language Processing.

Bio

Dr. José Manuel Gómez-Pérez leads the Cogito Research Lab at Expert System, where he focuses on the combination of neural and knowledge-based approaches to enable reading comprehension in machines. His work lies at the intersection of several areas of artificial intelligence, including natural language processing, knowledge graphs and deep learning. He also consults for organizations like the European Space Agency. A former Marie Curie fellow, José Manuel holds a Ph.D. in Computer Science and Artificial Intelligence from Universidad Politécnica de Madrid. He regularly publishes in top scientific conferences and journals and his views have appeared in magazines like Nature and Scientific American or national newspapers like El País.