Named entity recognition systems achieve remarkable performance on domains such as English news. It is natural to ask: What are these models actually learning to achieve this? Are they merely memorizing the names themselves? Or are they capable of interpreting the text and inferring the correct entity type from the linguistic context? We examine these questions by contrasting the performance of several variants of architectures for named entity recognition, with some provided only representations of the context as features. We experiment with GloVe-based BiLSTM-CRF as well as BERT. We find that context does influence predictions, but the main factor driving high performance is learning the named tokens themselves. Furthermore, we find that BERT is not always better at recognizing predictive contexts compared to a BiLSTM-CRF model. We enlist human annotators to evaluate the feasibility of inferring entity types from context alone and find that humans are also mostly unable to infer entity types for the majority of examples on which the context-only system made errors. However, there is room for improvement: A system should be able to recognize any named entity in a predictive context correctly and our experiments indicate that current systems may be improved by such capability. Our human study also revealed that systems and humans do not always learn the same contextual clues, and context-only systems are sometimes correct even when humans fail to recognize the entity type from the context. Finally, we find that one issue contributing to model errors is the use of “entangled” representations that encode both contextual and local token information into a single vector, which can obscure clues. Our results suggest that designing models that explicitly operate over representations of local inputs and context, respectively, may in some cases improve performance. In light of these and related findings, we highlight directions for future work.

The most widely adopted approaches for evaluation of summary content follow some protocol for comparing a summary with gold-standard human summaries, which are traditionally called model summaries. This evaluation paradigm falls short when human summaries are not available and becomes less accurate when only a single model is available. We propose three novel evaluation techniques. Two of them are model-free and do not rely on a gold standard for the assessment. The third technique improves standard automatic evaluations by expanding the set of available model summaries with chosen system summaries. We show that quantifying the similarity between the source text and its summary with appropriately chosen measures produces summary scores which replicate human assessments accurately. We also explore ways of increasing evaluation quality when only one human model summary is available as a gold standard. We introduce pseudomodels, which are system summaries deemed to contain good content according to automatic evaluation. Combining the pseudomodels with the single human model to form the gold-standard leads to higher correlations with human judgments compared to using only the one available model. Finally, we explore the feasibility of another measure—similarity between a system summary and the pool of all other system summaries for the same input. This method of comparison with the consensus of systems produces impressively accurate rankings of system summaries, achieving correlation with human rankings above 0.9.

Although there has been much theoretical work on using various information status distinctions to explain the form of references in written text, there have been few studies that attempt to automatically learn these distinctions for generating references in the context of computer-regenerated text. In this article, we present a model for generating references to people in news summaries that incorporates insights from both theory and a corpus analysis of human written summaries. In particular, our model captures how two properties of a person referred to in the summary—familiarity to the reader and global salience in the news story—affect the content and form of the initial reference to that person in a summary. We demonstrate that these two distinctions can be learned from a typical input for multi-document summarization and that they can be used to make regeneration decisions that improve the quality of extractive summaries.