CANTM

Implementation of Classification Aware Neural Topic Model (CANTM) and application for misinformation category classification Please cite the following paper in your publication involving this work: Xingyi Song, Johann Petrak, Ye Jiang, Iknoor Singh, Diana Maynard, Kalina Bontcheva, Classification Aware Neural Topic Model and its Application on a New COVID-19 Disinformation Corpus, In arXiv:2006.03354, 2020.

Install

• Requirement:
  • miniconda: https://docs.conda.io/en/latest/miniconda.html
• Install conda environment: conda env create -f environmentGPU.yml
• Active conda environment: conda activate wvmisinfoGPU
• Download NLTK and BERT models: python getPerpare.py

Run Experiments:

Covid Exp

• Get COVID data:

cd wvCovidData
bash getCovidData.sh

• Run CANTM: bash runCANTM_covid.sh
• Run BERT: bash runBert_covid.sh
• Run NVDM: runNVDM_covid.sh
• Run NVDM_bert: runNVDM_bert_covid.sh
• Run LDA: runLDA_covid.sh

Update topics with unlabelled data:

• Update classification-aware topic on COVID (note require `bash runCANTM_covid.sh' first), the latest unlabelled data not included please contact author. for the data: bash updateCovid.sh

Run Covid data on SCHOLAR:

• Perpare data for SCHOLAR: bash outputForScholar.sh
• The Scholar ready data format (splited in 5 folds) will be in 'testScholarfold'
• Please follow the instruction in Scholar to run the experiment
• Patched version of 'run_scholar.py' in Scholar_patch/, that output f1 score. (Copy to original Scholar folder to use)