Skip to content
conftrace_
2017 IJCNLP IJCNLP 2017

Leveraging Discourse Information Effectively for Authorship Attribution

Abstract

AbstractWe explore techniques to maximize the effectiveness of discourse information in the task of authorship attribution. We present a novel method to embed discourse features in a Convolutional Neural Network text classifier, which achieves a state-of-the-art result by a significant margin. We empirically investigate several featurization methods to understand the conditions under which discourse features contribute non-trivial performance gains, and analyze discourse embeddings.

🌉 Interdisciplinary Bridge - Artificial Intelligence and Natural Language Processing
🐣 Hot Topic Early Bird - feature embedding

Authors

Elisa Ferracane, Su Wang, Raymond Mooney

Topics

Machine Learning > Core Methods > Classification Deep Learning > Architectures > Neural Networks Natural Language Processing > Applications > Text Classification Machine Learning > Learning Types > Deep Learning

Keywords

feature learning text classification feature embedding authorship attribution convolutional neural network discourse analysis
Download PDF

More from IJCNLP 2017

All IJCNLP 2017 papers →
Capturing Long-range Contextual Dependencies with Memory-enhanced Conditional Random Fields
Named Entity Recognition with Stack Residual LSTM and Trainable Bias Decoding
Neuramanteau: A Neural Network Ensemble Model for Lexical Blends
Lightly-Supervised Modeling of Argument Persuasiveness
Multi-Task Learning for Speaker-Role Adaptation in Neural Conversation Models
Chat Disentanglement: Identifying Semantic Reply Relationships with Random Forests and Recurrent Neural Networks