WikiReading: A Novel Large-scale Language Understanding Task over Wikipedia
This work addresses a large-scale language understanding task for researchers in NLP and AI, providing a new benchmark for evaluating models, but it is incremental as it builds on existing methods for classification and extraction.
The authors tackled the problem of predicting textual values from Wikidata by reading Wikipedia articles, introducing the WikiReading dataset with 18 million instances, and found that a word-level sequence-to-sequence model with a copy mechanism achieved 71.8% accuracy.
We present WikiReading, a large-scale natural language understanding task and publicly-available dataset with 18 million instances. The task is to predict textual values from the structured knowledge base Wikidata by reading the text of the corresponding Wikipedia articles. The task contains a rich variety of challenging classification and extraction sub-tasks, making it well-suited for end-to-end models such as deep neural networks (DNNs). We compare various state-of-the-art DNN-based architectures for document classification, information extraction, and question answering. We find that models supporting a rich answer space, such as word or character sequences, perform best. Our best-performing model, a word-level sequence to sequence model with a mechanism to copy out-of-vocabulary words, obtains an accuracy of 71.8%.