LSDSem 2017: Exploring Data Generation Methods for the Story Cloze Test (2017-04-01T00:00:00.000000Z)

TL;DR

This work presents a system based on a deep learning architecture combined with a rich set of manually-crafted linguistic features that outperforms all known baselines for the Story Cloze test, suggesting that the chosen approach is promising.

Abstract

The Story Cloze test is a recent effort in providing a common test scenario for text understanding systems. As part of the LSDSem 2017 shared task, we present a system based on a deep learning architecture combined with a rich set of manually-crafted linguistic features. The system outperforms all known baselines for the task, suggesting that the chosen approach is promising. We additionally present two methods for generating further training data based on stories from the ROCStories corpus.

Authors

Iryna Gurevych

19 papers

Maxime Peyrard

4 papers

M. Bugert

3 papers

LSDSem 2017: Exploring Data Generation Methods for the Story Cloze Test

TL;DR

Abstract

Authors

References20 items

Improved Representation Learning for Question Answer Matching

A Corpus and Cloze Evaluation for Deeper Understanding of Commonsense Stories

Attentive Pooling Networks

LSTM-based Deep Learning Models for non-factoid answer selection

Applying deep learning to answer selection: A study and an open task

Adam: A Method for Stochastic Optimization

GloVe: Global Vectors for Word Representation

Text Rewriting Improves Semantic Role Labeling

DKPro TC: A Java-based Framework for Supervised Learning Experiments on Textual Data

Recursive Deep Models for Semantic Compositionality Over a Sentiment Treebank

Semi-Supervised Semantic Role Labeling via Structural Alignment

Deep Sparse Rectifier Neural Networks

Expanding textual entailment corpora fromWikipedia using co-training

Word Sense Disambiguation with Semi-Supervised Learning

Bidirectional recurrent neural networks

Long Short-Term Memory

Scripts, plans, goals and understanding: an inquiry into human knowledge structures

This Paper Is Included in the Proceedings of the 12th Usenix Symposium on Operating Systems Design and Implementation (osdi '16). Tensorflow: a System for Large-scale Machine Learning Tensorflow: a System for Large-scale Machine Learning

The systems previously explained in Section 3.3

To assess the quality of the KDE sampling method in isolation, we also include a BiLSTM system in the evaluation which is trained only on the ROCStories corpus with

Field of Study