- [DONE] Write tests for the
extract_aspects
functions and make sure everything runs as expected. - [DONE] Write sample for
extract_aspects
- [DONE]Sketch out
score_aspect
functions.- [DONE-but minimal]Get a sentence-level sentiment analysis model going.
- Write
test_score_aspect
tests.
- Refactor code in
get_sentences_by_aspect
(inscore_aspect
file) and code inextract_aspects
(inmain
file) ==> redundancy here. main.py
now runs, but seems to be buggy (all aspect scores are 0.0). Write tests in order to figure out where the bug is.
- Improve sentiment analysis
- Demonstrate ML chops
- Need to do some manual tagging?
- Improve opinion sentence identification (part of aspect extraction).
Today:
- [DONE] Write
test_score_aspect
tests.- [DONE] ensure that
main.py
works.
- [DONE] ensure that
- Get improved sentiment analysis going:
- Incorporate several lexicon-based featurizers.
- train/evaluate on movie review dataset.