FREyA

FREyA is a Natural Language Interface which can be used to query ontologies in RDF/OWL format. It maps Natural Language Query such as 'what is the capital of England?' to SPARQL - which then when executed against an RDF/OWL repository returns an answer.

FREyA Workflow

The implementation of FREyA can be broken down into several steps (see Figure 1):                           

  • Identification  of  Potential  Ontology  Concepts  (POCs). POCs are derived from the syntactic parse tree, and refer to question terms which could be linked to an ontology concept. Syntactic parse tree is generated by Stanford Parser (Klein and Manning, 2002).  We use several heuristic rules in order to identify POCs. For example, each NP (noun phrase) or NN (noun) is identified as a POC. Also, if a noun phrase contains adjectives, these are considered POCs as well.
  • Identification and disambiguation of Ontology Concepts
  • Mapping POCs to OCs, and engaging the user if necessary.
  • Grouping OCs into triples (Subject, Predicate, Object) 
  • Generating SPARQL dynamically from the set of triples
  • Identification of the answer type and presenting the concise answer to the user. We implemented the consolidation algorithm which relies on the syntactic parsing in order to identify the question focus (what the question is about), and the Answer Type Identificator (WH-phrases such as How big, or Where).  The consolidation algorithm then merges these with the ontology based lookup in order to identify the answer type (see Figure 2). See Damljanovic et al. (2011) for more details.


Figure 1. FREyA Workflow                                                                                                           Figure 2. Identification of the Answer Type

                                                          




References

Dan Klein and Christopher D. Manning. Fast Exact Inference with a Factored Model for Natural Language Parsing. In Suzanna Becker, Sebastian Thrun, and Klaus Obermayer, editors, Advances in Neural Information Processing Systems 15 - Neural Information Processing Systems, NIPS 2002, pages 3{10. MIT Press, 2002. ISBN 0-262-02550-7. URL: http://books.nips.cc/papers/files/nips15/CS01.pdf.

D. Damljanovic, M. Agatonovic, H. Cunningham: Identi.cation of the Question Focus: Combining Syntactic Analysis and Ontology-based Lookup through the User Interaction. In Proceedings of the 7th Language Resources and Evaluation Conference (LREC'10), ELRA 2010, La Valletta, Malta, May 17-23, 2010.
Comments