DAMON

Decomposable Attention MOdel for Novelty (DAMON): A technique to identify redundant documents

This is the codebase used in the writing of the paper called: "Is Your Document Novel? Let Attention Guide You. An Attention-Based Model For Document Level Novelty Detection" which is published in Natural Langauge Engineering (NLE) journal by Cambridge University Press.

Requirements

Tensorflow
Keras
PyTorch
Matplotlib
GLoVe common crawl pre-trained word vectors (http://nlp.stanford.edu/data/glove.840B.300d.zip)
TAP-DLND-1.0 (https://www.iitp.ac.in/~ai-nlp-ml/resources.html#)

dlnd: Contains various scripts and the DLND dataset in .zip format
- scripts: Contains various scripts used for obtaining results in the paper
apwsj: Contains .tar file containing judgements on APWSJ dataset and other .ttxt files derived from it
- scripts: Contains various scripts used for obtaining results in the paper
sentence_encoder: Contains scripts from Infersent used for generating sentence embeddings

Name		Name	Last commit message	Last commit date
Latest commit History 7 Commits
apwsj		apwsj
dlnd		dlnd
sentence_encoder		sentence_encoder
README.md		README.md