• Publications
  • Influence
Hierarchical Attention Networks for Document Classification
We propose a hierarchical attention network for document classification. Our model has two distinctive characteristics: (i) it has a hierarchical structure that mirrors the hierarchical structure ofExpand
  • 2,237
  • 362
  • PDF
Automatic Evaluation of Summaries Using N-gram Co-occurrence Statistics
TLDR
We show that automatic evaluation using unigram co-occurrences between summary pairs correlates surprising well with human evaluations based on various statistical metrics; while direct application of the BLEU evaluation procedure does not always give good results. Expand
  • 1,485
  • 215
  • PDF
End-to-end Sequence Labeling via Bi-directional LSTM-CNNs-CRF
TLDR
In this paper, we introduce a novel neutral network architecture that benefits from both word- and character-level representations automatically, by using combination of bidirectional LSTM, CNN and CRF. Expand
  • 1,387
  • 186
  • PDF
OntoNotes: The 90% Solution
TLDR
We describe the OntoNotes methodology and its result, a large multilingual richly-annotated corpus constructed at 90% interannotator agreement. Expand
  • 751
  • 147
  • PDF
RACE: Large-scale ReAding Comprehension Dataset From Examinations
TLDR
We present RACE, a new dataset for benchmark evaluation of methods in the reading comprehension task. Expand
  • 345
  • 103
  • PDF
Determining the Sentiment of Opinions
TLDR
We present a system that, given a topic, automatically finds the people who hold opinions about that topic and the sentiment of each opinion. Expand
  • 1,490
  • 67
  • PDF
Learning surface text patterns for a Question Answering System
TLDR
In this paper we explore the power of surface text patterns for open-domain question answering systems. Expand
  • 887
  • 52
  • PDF
Government 2.0: Making connections between citizens, data and government
TLDR
We introduce the concept of “open government,” so-called Government 2.0, and its required principles, functions and technological enablers to lead to a transformative, participatory model of e-government. Expand
  • 405
  • 43
  • PDF
Learning Whom to Trust with MACE
TLDR
We build a generative model of the annotation process that learns in an unsupervised fashion to identify which annotators are trustworthy and predict the correct underlying labels. Expand
  • 181
  • 39
  • PDF
Self-Training With Noisy Student Improves ImageNet Classification
TLDR
We present a simple self-training method that achieves 88.4% top-1 accuracy on ImageNet, which is 2.0% better than the previous state-of-the-art model that requires 3.5B weakly labeled Instagram images. Expand
  • 204
  • 36
  • PDF