-
pip install -r requirements.txt
-
install Matlab Python engine
-
git clone git@gitlab.com:tensors/tensor_toolbox.git
-
load_data.py
loads raw text as a N by D index matrix -
doc2vec.py
calculates tensor embeddings for each sentence -
label_progagation.py
calculates labels with label propagation -
tensor_decomp_twitter.py
calculates lexical centrality
-
download SemEval 2017 Task dataset, and put it into
./data
-
python tensor_decomp_twitter.py
-
python taskb_eval_script.py
-
put
16000 one liners
andpun of the day
datasets into./data
-
python cv_portion.py --option <option> --label_portion <label_portion>
, where<option>
can be16000oneliners
orPun
(corresponding to the16000 one liners
andpun of the day
datasets, respectively),<label_portion>
is a float number of training percentage, such as 0.1
- the dataset paths are configurated in
config.py
GNU GENERAL PUBLIC LICENSE