Navegación

Búsqueda

Búsqueda avanzada

Tuning Fuzzy SPARQL Queries in a Fuzzy Logic Programming Environment

We have recently designed FSA-SPARQL, an extension of the SPARQL query language for querying fuzzy RDF datasets. Answers of FSA-SPARQL queries are usually annotated with truth degrees which are computed from fuzzy connectives and operators that act on truth degrees associated to RDF triples. While FSA-SPARQL offers a rich repertoire of fuzzy connectives and operators, it is not always easy to retrieve the user’s expected answers. This is very often due to wrong formulation of queries, caused by inadequate use/combination of fuzzy connectives, operators and thresholds. For instance, a high threshold for truth degrees in some RDF datasets can lead to an empty set of answers, some strong or weak restrictive combination of fuzzy conditions might produce few or too many answers, etc. On the other hand, our research group has also developed the fuzzy logic programming language FASILL, which has been equipped with tuning techniques for enabling the customization of queries from test cases. In this paper, our goals are: (1) to provide a FSA-SPARQL translation to FASILL and (2) apply the tuning techniques to FSA-SPARQL queries for getting more precise formulation of queries from test cases.Artículo pendiente de publicación en el 2019 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE 2019): http://sites.ieee.org/fuzzieee-2019/

Ontology and Constraint Reasoning Based Analysis of SPARQL Queries

The discovery and diagnosis of wrong queries in database query languages have gained more attention in recent years. While for imperative languages well-known and mature debugging tools exist, the case of database query languages has traditionally attracted less attention. SPARQL is a database query language proposed for the retrieval of information in Semantic Web resources. RDF and OWL are standardized formats for representing Semantic Web information, and SPARQL acts on RDF/OWL resources allowing to retrieve answers of user’s queries. In spite of the SPARQL apparent simplicity, the number of mistakes a user can make in queries can be high and their detection, localization, and correction can be difficult to carry out. Wrong queries have as consequence most of the times empty answers, but also wrong and missing (expected but not found) answers. In this paper we present two ontology and constraint reasoning based methods for the discovery and diagnosis of wrong queries in SPARQL. The first method is used for detecting wrongly typed and inconsistent queries. The second method is used for detecting mismatching between user intention and queries, reporting incomplete, faulty queries as well as counterexamples. We formally define the above concepts and a batch of examples to illustrate the methods is shown.

Fuzzy Queries of Social Networks involving Sentiment Analysis and Topic Detection (Trabajo en progreso)

Social networks have become a source of data which are of interest in all areas, and their querying and analysis is a hot topic in computer science. Our research group has developed a fuzzy extension of the Semantic Web query language SPARQL, called FSA-SPARQL (Fuzzy Sets and Aggregators based SPARQL). This extension provides mechanisms to express fuzzy queries against RDF data. FSA-SPARQL works with social networks. With this aim, FSA-SPARQL enables the transformation and fuzzification of social network API data. Fuzzification of social networks data is automatic and user-defined enabling a wide range of mechanisms for ranking and categorization, including sentiment analysis and topic detection. As case study, FSA-SPARQL has been used to query three well-known social networks: Twitter, Foursquare and TMDb.

iHDT++: un Autoíndice Semántico para la Resolución de Patrones de Consulta SPARQL

La publicación de colecciones RDF, y el volumen de las mismas, ha crecido exponencialmente en los últimos años, abriendo nuevos retos de investigación relacionados con el almacenamiento, el procesamiento y la consulta de Big Semantic Data. Los auto-índices RDF son una de las soluciones más innovadoras en este escenario, ya que no sólo comprimen las colecciones, sino que además proveen acceso eficiente a los datos sin descomprimirlos previamente. En este escenario, HDT es una de las soluciones de referencia y su uso ha sido validado por diferentes herramientas semánticas. Sin embargo, la efectividad de HDT está limitada por la sencillez de su diseño y sus ratios de compresión han sido recientemente mejorados por HDT++. Sin embargo, HDT++ no soporta directamente la resolución de consultas SPARQL. En este artículo extendemos HDT++ para dar soporte a la resolución de todos los triple patterns SPARQL. Esta nueva propuesta (iHDT++) mejora los resultados de compresión obtenidos por HDT y garantiza un rendimiento comparable para la resolución de consultas.

FSA-SPARQL: Fuzzy Queries in SPARQL (Trabajo en progreso)

SPARQL has been adopted as query language for the Semantic Web. RDF and OWL have been also established as vocabularies to describe ontologies in this setting. While RDF/OWL/SPARQL have been designed for querying crisp information, some contexts require to manage uncertainty, vagueness and imprecise knowledge. In this paper we propose a SPARQL extension, called FSA-SPARQL (Fuzzy Sets and Aggregators based SPARQL) in which queries can involve different fuzzy connectives and (aggregation) operators. The language has been implemented as an extension of the ARQ Jena SPARQL engine and it is equipped with a Web tool from which queries can be executed on-line.

Type Checking and Testing of SPARQL Queries (Trabajo en progreso)

In this paper we describe a property-based testing tool for SPARQL. The tool randomly generates test cases in the form of instances of an ontology. The tool checks the well typed-ness of the SPARQL query as well as the consistency of the test cases with the ontology axioms. With this aim, a type system has been defined for SPARQL. Test cases are after used to execute queries. The output of the queries are tested with a Boolean property which is defined in terms of membership of ontology individuals to classes. The testing tool reports counterexamples when the Boolean property is not satisfied.