Before saving spotted, check its spelling and change its words
Save changed words as a dictionary with "new_word" : "previous" so that they can be put back if needed
Use this model http://norvig.com/spell-correct.html
To create the predictor. Use the whole dataset to create the bag of words
Create tokenizer to fix slangs, things like "vc" and repeated letters in words. Also try to remove urls and such
Experiment with priority queues and binary trees to make it faster.
Experiment saving common mistakes to fasten it up
Before saving spotted, check its spelling and change its words
Save changed words as a dictionary with "new_word" : "previous" so that they can be put back if needed
Use this model http://norvig.com/spell-correct.html
To create the predictor. Use the whole dataset to create the bag of words
Create tokenizer to fix slangs, things like "vc" and repeated letters in words. Also try to remove urls and such
Experiment with priority queues and binary trees to make it faster.
Experiment saving common mistakes to fasten it up