This repo contains code which classifies Sanskrit slokas into 3 categories: Chanakya Slokas, Vidur Neeti Slokas and rest of the Slokas as Sanskrit Slokas
Here is the link to the web app. click here
(web app server is currently down)
We applied LSTM network to the attached datasets and found the accuracy to about 85 %
To improve this further and make it usable for general public, we have thought of the following improvements:
- Improving the dataset. We will have to add more more examples and annotate the given Sanskrit Slokas into more classes.
- Use better models example BERT and transformers and use the one with the best accuracy.
- Implement a Sanskrit lemmatizer to enhance the use of Data.
If you would like to contribute to any of the above, kindly create a pull request in a different branch and mention @kamalojasv181 or @Aparna-Sakshi in the pull request.
If you would like to improve it in other ways, kindly do not submit a pull request directly, rather create an issue and mention @kamalojasv181 or @Aparna-Sakshi.