Vec2SPARQL: integrating SPARQL queries and knowledge graph embeddings

<div>Recent developments in machine learning have led to a rise of large</div><div>number of methods for extracting features from structured data. The features</div><div>are represented as vectors and may encode for some semantic aspects of data.</div><div>They can be used in a machine learning models for different tasks or to com-</div><div>pute similarities between the entities of the data. SPARQL is a query language</div><div>for structured data originally developed for querying Resource Description Frame-</div><div>work (RDF) data. It has been in use for over a decade as a standardized NoSQL</div><div>query language. Many different tools have been developed to enable data shar-</div><div>ing with SPARQL. For example, SPARQL endpoints make your data interopera-</div><div>ble and available to the world. SPARQL queries can be executed across multi-</div><div>ple endpoints. We have developed a Vec2SPARQL, which is a general frame-</div><div>work for integrating structured data and their vector space representations.</div><div>Vec2SPARQL allows jointly querying vector functions such as computing sim-</div><div>ilarities (cosine, correlations) or classifications with machine learning models</div><div>within a single SPARQL query. We demonstrate applications of our approach</div><div>for biomedical and clinical use cases. Our source code is freely available at</div><div>https://github.com/bio-ontology-research-group/vec2sparql and we make a</div><div>Vec2SPARQL endpoint available at http://sparql.bio2vec.net/</div>