5% FOR CORRECT UNSUPPORTED ANSWERS. ALSO, THE ACCURACY OF IMPROVEME...

44.5% for correct unsupported answers. Also, the accuracy of improvement of 92%. their question classification module was 88% using InfoMap Limitations and 73.5% using SVM. The model was tested in a restricted experimental environment in which the domain was very specific and the  Stoyanchev et al. (StoQA) [6], 2008: number of questions is relatively small. Also, depending only Contribution on the users as a single source for validating answers is a two-In their research, they presented a document retrieval edged weapon. experiment on a question answering system. They used exact phrases, which were automatically identified from questions,