Who did What: A Large-Scale Person-Centered Cloze Dataset
This provides a new benchmark for the NLP community to evaluate reading comprehension systems, though it is incremental as it builds on existing cloze-style datasets.
The authors tackled the need for a more challenging reading comprehension dataset by constructing the 'Who-did-What' dataset with over 200,000 cloze problems from news articles, avoiding article summaries and anonymization, and filtering to ensure 84% human solvability.
We have constructed a new "Who-did-What" dataset of over 200,000 fill-in-the-gap (cloze) multiple choice reading comprehension problems constructed from the LDC English Gigaword newswire corpus. The WDW dataset has a variety of novel features. First, in contrast with the CNN and Daily Mail datasets (Hermann et al., 2015) we avoid using article summaries for question formation. Instead, each problem is formed from two independent articles --- an article given as the passage to be read and a separate article on the same events used to form the question. Second, we avoid anonymization --- each choice is a person named entity. Third, the problems have been filtered to remove a fraction that are easily solved by simple baselines, while remaining 84% solvable by humans. We report performance benchmarks of standard systems and propose the WDW dataset as a challenge task for the community.