Querying Misaligned Data Sources

Experiments used in the paper's main text and appendix.

Setup

python -m pip install -r requirements.txt

1. Attribute scalability

python -m experiments.cartesian_attributes.run_all

python -m experiments.cartesian_attributes.run_all \
  --prior-model gaussian-random \
  --strategies alg1

python -m experiments.cartesian_attributes.run_distributional

2. Trustworthy-answer detection

python -m experiments.trustworthy_answers.run_scalability

3. MI bucketization

python -m experiments.mi_bucketization.run_all

4. Conflict learning

python -m experiments.conflict_learning.run_all

Tests

python -m pytest

Results and generated figures are in results/. The Amazon dataset is too large to include in this repository. Download it from the Amazon Products Dataset on Kaggle and place amazon_products.csv in data/real/.