I am currently exploring and playing with wide NLP techniques and projects, and one project idea that stumbled on my mind and gave me the intuition that it can lead to great learning is to build a small Grammarly clone, even with only spell-checking. So if anyone has experience doing similar projects, please describe the main steps and ideas for implementation.
This is interesting. I have very deep interest in NLP and linguistics. I would really like to follow on your research.
I would start initially with building parser of a sentence. For example, parsing a sentence to identify correct structure (subject + verb + object) and other structures. Once done, I would try to detect error, and then correct form. Third step would be to check for typos and spellings. Fourth step would be readability checking. Fifth step could be tone checking. Sixth step would be, suggestion for better readability. Seventh step could next word prediction.
There are several possibilities IMO. If you are looking for backend, then I would also think about technological frame. I would start with django, or flask as they are python based and there are more tools for NLP in python than any other programming language.
- Badges
- Science topic