We have finished our first internship between InSTEDD and Trinity College. Juan Pablo and Qianqian did an amazing job putting together an SVM classifier using libSVM for suggesting tags inside Riff items. They tested the implementation with Reuters dataset and a Promed dataset that we created using Riff.
They ran 4 independent tests to find values for the specificity and sensitivity of libsvm against Reuters and Promed.
They've also worked in a Naive Bayes classifier to suggest tags based on previous evidence of symptoms relationships. You can find here a description of their work.
Thanks Qianqian, Juan Pablo, Trishan and everybody at Trinity College, it was a pleasure to work with you, I hope we meet again soon.