A semantic system for answering questions in neuroinformatics

Neuroinformatics is an important area of study in biomedical science and health informatics. Scientists in neuroscience tend to ask questions that are complicated, time consuming to answer and need multiple tasks in order to get to the result. In this paper, we introduce and report an ontology-based system for answering neuroscience questions automatically. The system uses a combination of ontologies such as NIFSTD and NeuroFMA, a template-based question translation method that translates questions to SparQL codes and MRI outputes (annotations) in order to classify and answer questions. It also uses an ontology-based query expansion module. The outcomes show the ontology-based question classification achieves 87.5% correct classification on the data set and the system can successfully answer 78.13% of questions. This research also uses machine learning techniques such as Naïve-Bayes, KNN, SVM and Random Forest to classify questions which respectively result in 54.54%, 68.18%, 72.72% and 77.27% correct classification.