Social media scraper that looks through user profiles for profane or obscene messages. The identification mechanism utilizes an NLP algorithm trained on a Kaggle dataset.
- Make frontend aesthetically pleasing
- Connect frontend w/ backend POST
Backend
- the twitAPI folder contains classes that interface to twitter to gather the necessary data
- Server resources folder countains the files that actually create objects and call the methods, before sending them to the server.
WITHIN TWITAPI:
-Twitterer.java:
- This was the original tweet grabbing function, but it's limited and should BE PHASED OUT.
- Twitterfunctions still uses it for some reason.
- Twitterfunctions.java:
- /b calls two classes to grab tweets; twitterer and twitter. I have no idea where twitter is from, but twitterer is from Twitterer.Java
- Checks if a profile is highschool by checking the bio for a height statement, or a grad year
- These are implemented with manual checks for years/height
- Looking ahead, we need a better method of doing this.
- Checks a list of handles for highschoolers using the above function.
- Has a method to grab URLs
- Has a method to grab the profile name
- Checks for interested universities using a keyword search for things like "blessed", "receive" etc.
- Has the method which uses micheal's NLP thingy to find flagged tweets for a handle. ()
- Has a method to check rleative interest based on the amoun of GT tweets retweeted (find better data to use for this eventually)
- percentfit method, currently incomplete but will be based off of evaluations like the one above in the future.
- maybe look at a different model to express fit. a percent may not be the best.
- TwitterDriver.java:
- This might be completely obsolete but I'll have to test it to make sure
- Essentially just calls twitterer and prints some tweets to console. Pretty sure I just created it and messed with it a bit to check out functions of TwitAPI
William Chen, Margarita Groisman, Michael Chen, Jeffrey Luo