• Publications
  • Influence
DORA The Explorer: Directed Outreaching Reinforcement Action-Selection
TLDR
We propose $E$-values, a generalization of counters that can be used to evaluate the propagating exploratory value over state-action trajectories. Expand
  • 27
  • 8
  • PDF
SemEval 2019 Task 1: Cross-lingual Semantic Parsing with UCCA
TLDR
We present the SemEval 2019 shared task on Universal Conceptual Cognitive Annotation (UCCA) parsing in English, German and French, and discuss the participating systems and results. Expand
  • 23
  • 5
  • PDF
Will it Blend? Blending Weak and Strong Labeled Data in a Neural Network for Argumentation Mining
TLDR
We propose a methodology to blend high quality but scarce strong labeled data with noisy but abundant weak labeled data during the training of neural networks, and show the advantages of the blending scheme. Expand
  • 19
  • 4
  • PDF
Inherent Biases in Reference-based Evaluation for Grammatical Error Correction and Text Simplification
TLDR
We show that overcoming LCB in Grammatical Error Correction (GEC) evaluation cannot be attained by re-scaling or by increasing the number of references in any feasible range, contrary to previous suggestions. Expand
  • 13
  • 1
  • PDF
Reference-less Measure of Faithfulness for Grammatical Error Correction
TLDR
We propose USim, a semantic measure for Grammatical Error Correction (that measures the semantic faithfulness of the output to the source, thereby complementing existing reference-less measures (RLMs) for measuring the output's grammaticality. Expand
  • 22
  • PDF
Are You Convinced? Choosing the More Convincing Evidence with a Siamese Network
TLDR
We present a new data set, IBM-EviConv, of pairs of evidence labeled for convincingness, designed to be more challenging than existing alternatives. Expand
  • 14
  • PDF
Corpus Wide Argument Mining - a Working Solution
TLDR
We present a first end-to-end high-precision, corpus-wide argument mining system by combining sentence-level queries over a very large corpus of newspaper articles, with an iterative annotation scheme. Expand
  • 9
  • PDF
Learning to combine Grammatical Error Corrections
TLDR
We propose a new method for combining multiple GEC systems, using a pure black-box approach, can improve state of the art results in the error correction task. Expand
  • 9
  • PDF
On the Weaknesses of Reinforcement Learning for Neural Machine Translation
TLDR
We prove that one of the most common RL methods for MT does not optimize the expected reward, as well as show that other methods take an infeasibly long time to converge. Expand
  • 9
  • PDF
Automatic Metric Validation for Grammatical Error Correction
TLDR
We propose MAEGE, an automatic methodology for GEC metric validation, that overcomes many of the difficulties with existing practices. Expand
  • 8
  • PDF