I'm working on a project to do sentiment analysis but my data is not long and properly formatted text. It's more likely to be very short sentences, e.g. tweets (in full tweet lingo), quick reviews of maybe 2-5 short sentences, etc.
If my text is of that nature, what approach would you recommend? E.g. CNNs (spaCy has a ready-made text classifier), LSTM (e.g. something like Keras), etc.
What are the pros/cons of your suggested approach (i.e. why is it better suited for classifying short paragraphs/sentences)?
I'm starting out in the area so any links/papers/etc. will be most welcome!