Business AI Tops the SQuAD Leader Board

Machine Reading Comprehension (MRC) is a relatively new part of natural language processing (NLP) and is an important part of AI that researchers around the world are working on. The MRC research community has been collaborating around several recently proposed datasets, e.g., the Stanford Question Answer Dataset (SQuAD) and the CNN/Daily Mail QA (CNN-DM) which was released by Google's DeepMind. The dataset generating the most activity in the research community since it was unveiled in 2016 is SQuAD. If you are not familiar with SQuAD, according to their website it is the "reading comprehension dataset, consisting of questions posed by crowd workers on a set of Wikipedia articles, where the answer to every question is a segment of text, or span, from the corresponding reading passage." One of the features of SQuAD is that they maintain a leaderboard where we can track the rapid progress that is being made in MRC and see how our latest models are doing relative to the state of the art.