search

CraigBryan / tweet-mood-analyzer

Assignment 2 for CSI 4107
0 stars 1 forks source link

Using emoticons #5

Open CraigBryan opened 9 years ago

CraigBryan commented 9 years ago

It would be neat if we could mark and categorize emoticons into the mood they portray. This would be a strong indicator, as it corresponds very strongly to the mood of a tweet.

BonShillings commented 9 years ago

Ya this would be a great feature definitely. I'll look into an emoticon dictionary, even if only a limited one to extract like :) :D and :( :@

CraigBryan commented 9 years ago

It would be wise to look into the various codes for common ascii/unicode symbols, like the 'heart' (I've seen it a couple times)

BonShillings commented 9 years ago

Im going to take a look into this tonight. I have to do another assignment today first, but Ill start looking into it after I finish, which may be in csi 4107 today

CraigBryan commented 9 years ago

When this is implemented, it should be done as a totally different filter that re-processes the original tweet string. The other filters do a lot of operations that will mangle the emoticons.

I think it'll be wise to just count positive and negative emoticons and have those two numbers as the features.

I'll start compiling a list of positive and negative emoticons in constants.py.