/DAEM

Deep Entity Matching with Adversarial Active Learning

Primary LanguagePython

DAEM

Source code, datasets and experimental results for submission "Deep Entity Matching with Adversarial Active Learning".

Dependencies

  • Python 3.8
  • Python libraries: see requirements.txt
  • Dataset and pre-trained word embeddings: please use download.sh to download

Datasets (from DeepMatcher)

We evaluate DAEM on seven benchmark datasets.

Project structure

FileDescription
daemCode of DAEM
scripts/entity_matching.pyCode to reproduce the deep entity matching experiment and ablation study
scripts/adversarial_active_learning.pyCode to reproduce the active entity matching experiment
scripts/inter_attribute_completion.pyCode to reproduce the inter-attribute completion experiment
scripts/adversarial_learning.pyCode to reproduce the adversarial learning experiment
scripts/entity_signature.pyCode to reproduce the entity signature experiment
scripts/dynamic_blocking.pyCode to reproduce the dynamic blocking experiment