Supervised aspect extraction

[alexanderpanchenko]

Step 1. Prepare a dataset with the OTE annotations in the CoNLL format. SemEval data:

Paper:
http://www.aclweb.org/anthology/S16-1174

Website:
http://alt.qcri.org/semeval2016/task5/

ABSA dataset:
https://drive.google.com/open?id=1TBkw8M53tqqV6v_7JHQCn6qq5l6CAhEV

CoNLL format:
http://universaldependencies.org/format.html

Step 2. Train a classifier to detect OTE in text

Tagger:
https://github.com/achernodub/bilstm-cnn-crf-tagger

For the beginning just try running example datasets from this repository. Then format the SemEval ABSA dataset in this format and train a model.

[ChulioZ]

Could you tell me the name of the Python package I should use for training the models etc.?

[alexanderpanchenko]

https://github.com/achernodub/bilstm-cnn-crf-tagger

[alexanderpanchenko]

if you need to install with pip:

pip install git+
https://github.com/achernodub/bilstm-cnn-crf-tagger

[alexanderpanchenko]

about the tagger and its model:
_2018_Grammarly_v15_preFinal_05.pdf

[ChulioZ]

Doesn't seem to fit our project. Closed for now.