Paper: Qi Jia, Mengxue Zhang, Shengyao Zhang, Kenny Q. Zhu: Matching Questions and Answers in Dialogues from Online Forums. ECAI 2020: 2046-2053

• tensorflow>=1.0
• numpy

Download the session-level and pair-level dataset from Google Drive and unzip the file under ./datafile.

The files includes:

• The annotated dialogues which are split into train/dev/test sets by 7:1:2
  • train-700.json
  • dev-100.json
  • test-200.json
• The reconstructed Q-NQ pairs used for our proposed model:
  • train-full.json
  • dev-full.json
  • test-full.json
• Pretrained word embeddings:
  • word_emb_reduce.txt
  • word2idx-new.json

Here is the breakdown of the commands :

• Train the model with

python train.py

• Evaluate and interact with the model with (config.py batch_size=1)

python evaluate.py

• Get the session predictions

python session_predictions.py
python outputPredict_True_role_sen.py