BiLSTM-Based Models for Metaphor Detection

This repository is the Lasagne implementation for the network presented in:

Shichao Sun and Zhipeng Xie, BiLSTM-Based Models for Metaphor Detection NLPCC 2017

Concat: [email protected]

Requirements

• python 2.7
• Lasagne 0.2.dev1
• gensim 1.0.1
• scikit-learn 0.18.1
• NLTK 3.2.2
• Theano 0.8.0
• Numpy 1.12.1

Train and Test

• Stage 1: Download the word2vec pre-trained Google News corpus GoogleNews-vectors-negative300.bin.gz
• Stage 2: run the code to train and test

python metaphor.py

Citation

@inproceedings{sun2017bilstm,
title={BiLSTM-Based Models for Metaphor Detection},
author={Sun, Shichao and Xie, Zhipeng},
booktitle={National CCF Conference on Natural Language Processing and Chinese Computing},
pages={431--442},
year={2017},
organization={Springer}
}