/ELJST-Paper

Primary LanguageJupyter Notebook

ELJST-Paper

Instructions to run a new dataset.

  • Make a copy of ELJST_Onetime and import the dataset and make the necessary preprocessings.
  • Use the same ipynb to generate the edges.
  • further, Test_ELJST_Onetime_only_edges.ipynb and Test_ELJST_Onetime_only_edges_attention_all.ipynb can be used to generate edges.
  • attention_all can be generated only in attention_all.ipynb.
  • all the edges will be in the resources folder.
  • baselines can be run using the nohookup (nosampler_uni, nosampler_etm (mrf-lda), for etm there is a different repo).
  • for lda, Baseline_LDA ipynb. Evaluations for LDA will be in the same file
  • once the models are dumped, evaluation scipts should be run seperately.
  • For MRF-LDA: Evaluation_ETM
  • For ELJST: ELJST-Evaluation
  • For ETM: Seperate Repo.

ETM Repo: https://github.com/williamscott701/ETM-1.git