This package is a basic sentiment analyzer
Sentipy
attempts to classify a text into positive
, negative
or neutral
sentiment. It uses Spacy's textcat
with ensemble architecture in the back-end.
The ultimate objective of this package is to classify the sentiments as
accurately as possible
While the sentiment analysis at the core is absolutely basic, the current focus
is to understand the features the model is learning. sentipy
leverages on
lime
to get the features learnt and uses streamlit
to crate a simple webapp
that helps with the visualization
- git clone the repo in to your local system
- run setup.py install
On your terminal run sentipy streamlit
- Use transformers and take the self supervised learning approach for classification
- Include visualizations for pre-processed text
- Make pre-processing options available on web app
- Better visualization
test changes made by me