Supportr is the code release for labeling replies on the relative
supportiveness, based on the paper It’s going to be okay: Measuring Access to Support in Online Communities by Zijian Wang and David Jurgens (in proceedings
of EMNLP 2018).
See the project website for full details, including contact information.
The code is not yet setup as a python package but the code can be run using the example.sh script in the main directory.
The setup.py file lists the model dependencies. Aside from these python
packages, the code requires the Google news word2vec vectors in the resources/
directory. The model described in the paper uses LIWC categories as features.
LIWC cannot be redistributed so these categories are not included in this release. However, the Empath
library is known to closely approximate the categories and the code is setup to
Just Work™ if you don't have LIWC purchase. If you do have LIWC, it should be
put in resources/lexicons/ and named en_liwc.txt. Additional tests are
needed to report performance when LIWC is not used.
Included in this release is the aggregrated and labeled training data for the paper, which can be used to reproduce the results of the paper or improve the classifier.
- 0.1 Initial code release
The next step is to package the code up as a module to allow easier classification. Pull requests welcome.