SPARQL2AL: Translating SPARQL Queries to Arabic Language

  • Omar Salah El Radie -----> Dr. Iyad Al-Agha

With the wide spread of Open Linked Data and Semantic Web technologies, a larger amount of data is published on the Web in the RDF and OWL format. This data can be queried and retrieved using the Semantic Web Query Language SPARQL. 

SPARQL is one of the three core technologies of the Semantic Web beside OWL and RDF. However, SPARQL cannot be understood by naïve users and it is not directly accessible to humans, and thus, they will not be able to check whether the retrieved answers truly correspond to the intended information needed. Because natural language generation from RDF data has recently become an important topic for research, this has led to the development of various systems generating natural language text from knowledge bases.

This work proposes an approach to translate SPARQL to the Arabic language.
It introduces the SPARQL2AL system that can interface to any Arabic ontology, get a SPARQL query as an input and retrieve valid Arabic statements. While few efforts proposed approaches to transform SPARQL queries to the natural language, most of these efforts focused on English domain. To convert SPARQL to Arabic natural language the SPARQL query passes through several phases; firstly, translations of query terms are extracted from the ontology, assuming that there exists an ontology of domain terms on which we run SPARQL queries to retrieve data. Secondly, we define
a set of Arabic language dependencies to identify how query terms are mapped to Arabic sentences. Thirdly, define a set of rules to transform parts of SPARQL to
Arabic sentences. Finally, we remove redundancies, group the results and output the Arabic statement.
The proposed system has been tested with a sample ontology and a query set consisting of 40 SPARQL queries. The accuracy of this system is 0.85