chao-jiang / iclr2017

Doc2VecC from the paper "Efficient Vector Representation for Documents through Corruption"

Doc2VecC

code from the paper Efficient Vector Representation for Documents Through Corruption.

Acknowledge

The code was modified from Thomas Mikolov's code on Paragraph Vector. https://groups.google.com/forum/#!msg/word2vec-toolkit/Q49FIrNOQRo/J6KG8mUj45sJ

Dependencies

You will need to download the liblinear package, and change the path to the package in the script accordingly. https://www.csie.ntu.edu.tw/~cjlin/liblinear/

Getting started

Run the script go.sh, it will download the IMDB movie review dataset, and learn document representations on this dataset. A linear SVM is trained on the learned representation fo sentiment analysis.

Reference

If you found this code useful, please cite the following paper:

Minmin Chen. "Efficient Vector Representation for Documents Through Corruption." 5th International Conference on Learning Representations, ICLR (2017).

@article{chen2017efficient,
  title={Efficient Vector Representation for Documents Through Corruption},
  author={Chen, Minmin},
  journal={5th International Conference on Learning Representations},
  year={2017}
}

License

Apache License 2.0