MUSKA: MUSic Knowledge Access

    • Natural Language Question Answering on Musical Scores: It aims to develop system that answers the natural language queries over the musical scores. The current approach is
      based upon two main steps: identifying the musical entities and relations present in the query, and retrieving the relevant music passages containing those entities from the associated MusicXML file. Musical entities in the query are recognized with the help of regular expressions. It initiated with the MediaEval 2014 Camerata Task. We submitted two runs for the task. The first one takes a union of the passages retrieved for each musical entity, while the second approach takes their intersection to answer the query.
      Please find the implementation here: GitHub repo

      Please find a short paper describing the implementation here: PDF

 

  • Musicology: We are developing ontologies to define the musical scores. This would help us to make interesting queries on the linked data created from the musical scores following the ontology.