Vec2SPARQL: integrating SPARQL queries and knowledge graph embeddings

Abstract

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

SGPT: A Generative Approach for SPARQL Query Generation From Natural Language Questions

“…The term "SPARQL" stands for SPARQL Protocol and RDF Query Language. According to the official definition, a SPARQL query can be formally considered as a tuple ⟨GP, DS, SM, R⟩, where GP is a graph pattern (query pattern), DS is an RDF dataset, SM is a set of solution modifiers (ORDER, PROJECTION, DISTINCT, OFFSET, LIMIT), R is a result form (SELECT, CONSTRUCT, DE-SCRIBE and ASK) 2 . Figure 2 illustrates the terms used in the formalization.…”

“…(1) SPARQL templates are usually created manually or semiautomatically by domain experts, which is both time consuming and cost intensive, (2) The query templates are tailored to a particular KG, which results in potentially changing of the whole template set when the underlying graph is changed, (3) The extension of template sets to handle new question types is performed manually or semi-automatically, and (4) In pipeline-based approaches, the SPARQL generation module is dependent on the performance of the preceding modules (i.e., entity and relation linkers as well as ranking algorithms) and, thus, suffer from error propagation.…”

Semantic similarity and machine learning with ontologies

“…These methods have in common that they use ontologies to solve biomedical problems that reside outside the domain of research on the ontologies themselves. However, there can be substantial methodological overlap with research on ontology engineering, ontology learning, quality control, querying, and reasoning with ontologies, as ontology embeddings can also be used for ontology alignment [ 139–141 ], as part of automated reasoning systems [ 142 , 143 ] or to query knowledge bases [ 80 , 144 ]. In the future, we expect to see even more integrated research on developing ontologies, ontology infrastructure and novel biomedical applications to which they can be applied.…”