Broad Context Language Modeling as Reading Comprehension
This work addresses the problem of broader context language modeling for researchers in NLP, representing an incremental advance by applying existing comprehension models to a specific dataset.
The paper tackled the LAMBADA word prediction task by framing it as a reading comprehension problem and applying neural network models, improving state-of-the-art accuracy from 7.3% to 49%. It analyzed 100 instances to identify strengths in handling dialogue cues and weaknesses in coreference resolution and external knowledge.
Progress in text understanding has been driven by large datasets that test particular capabilities, like recent datasets for reading comprehension (Hermann et al., 2015). We focus here on the LAMBADA dataset (Paperno et al., 2016), a word prediction task requiring broader context than the immediate sentence. We view LAMBADA as a reading comprehension problem and apply comprehension models based on neural networks. Though these models are constrained to choose a word from the context, they improve the state of the art on LAMBADA from 7.3% to 49%. We analyze 100 instances, finding that neural network readers perform well in cases that involve selecting a name from the context based on dialogue or discourse cues but struggle when coreference resolution or external knowledge is needed.