This is the official repository for our paper "ConQueR: Contextualized Query Reduction using Search Logs" in SIGIR'23.
Unfortunately, the search log dataset is not publicly available due to privacy issues. Instead, we provide sample data as an example of the data format.
conda env create -n qr python==3.7.7
conda activate qr
pip install -r requirements.txt
- Run the python file (at the root of the project)
# run conquer_core
python conquer_core.py
# run conquer_sub
python conquer_sub.py
- Run the python file (at the root of the project) with best_ckpt file of ConQueR_core and ConQueR_sub in saves folder.
# ensemble two models
python eval_ensemble.py
- gpus
- default: 0