Towards querying bioinformatic linked data in natural language

Even though Linked Data is a fairly young concept, querying its stores renews the old challenge of querying data in an easy and yet sufficiently expressive way. But, if in case of querying traditional relational databases, the knowledge of complete structure was feasible, in case of Linked Data it is far more difficult to have an exhaustive view. From the perspective of facilitating the access to the data to end-users with no prior expertise, the most natural solution would be querying in natural language. But, from the technological point of view, Natural Language Processing still has many limitations. In this context, we introduce our first results towards a system for building SPARQL queries from NL sentences that is based on sentences' grammatical structure, general structural patterns and ontological descriptions. We focus on processing queries from pharmacology data of Bio2RDF project.