#Tensorflow Twitter stances classification Use Tensorflow to run CNN, and RNN model. And add word2vec to represent the text. Ipython file is ready to run. Please check the following description for more detail. The original data can be found here: PHEME rumour scheme dataset: journalism use case, version 2
Each tweet in the tree-structured thread will have to be categorised into one of the following four categories
- Support
- Deny
- Query
- Comment
##1.CNN_word embedding
Requrement(Optional)
- Download full dataset "rumoureval-data"
- RUN:
python train.py
Method from IMPLEMENTING A CNN FOR TEXT CLASSIFICATION IN TENSORFLOW
My code is for competition RumourEval-2017, subtask A
##2.RNN_word2vec
Requrement(Optional)
- Download vocab.npy
- OR Downolad GloVe to build the vocab.npy data. This will take the program for about an hour.
- RUN:
python rnn_stance.py
GloVe for word2vec expression. LOO (Leave One Out) to evaluate the result. That is, use only one conversion thread as testing data. The rest data as training data.
Then we compare with other learning method on different rumor events. Performance especial good for F1 measure.
