|Thread title||Replies||Last modified|
|Feedback||0||17:29, 16 October 2017|
Your topic sounds interesting. Feel free to change the "something extra" part if you get stuck.
You may define a scoring function that reflects the semantic correctness of a sentence. Then you can use it to select the sentences with the highest score. Notice that my suggestion is totally optional.
Deciding "meaningfulness" is non-trivial. I look forward to seeing how ML can help you get a good model.