SwayambhuNathRay/AD3-EMNLP-2018

EMNLP 2018: AD3: Attentive Deep Document Dater

AD3

AD3: Attentive Deep Document Dater

Source code and dataset for EMNLP 2018 paper: AD3: Attentive Deep Document Dater.

Overview of AD3 (proposed method), an attention-based neural document dating system which utilizes both context and temporal information in documents in a flexible and principled manner. Please refer paper for more details.

Dependencies

• Compatible with TensorFlow 1.x and Python 3.x.
• Dependencies can be installed using requirements.txt.

Dataset:

• We evaluate AD3 on NYT and APW section of Gigaword Corpus, 5th ed. For preprocessing refer NeuralDater.

Usage:

• After installing python dependencies from requirements.txt, execute sh setup.sh for downloading GloVe embeddings.

• ac_gcn.py and oe_gcn.py contains TensorFlow (1.x) based implementation of AD3 (proposed method).

• To start training:

  python ac_gcn.py -data data/nyt_processed_data.pkl -class 10 -name test_run
  python oe_gcn.py -data data/nyt_processed_data.pkl -class 10 -name test_run

  • -class denotes the number of classes in datasets, 10 for NYT and 16 for APW.
  • -name is arbitrary name for the run.

NB:

• Source code can also be found at MALL Lab. This is just a copy.