This tool visualises the sentiment in twitch chat over time. It does so by using VADER extended with custom dictionaries to capture the sentiment of twitch emotes and other words unique to twitch chat.

In it's current state it only recognizes a subset of the default Twitch emotes and a subset of the emotes introduced by the BetterTTV extension. All other emotes (including channel specific ones) are considered 'neutral', unless specified in a dictionary file in vader_extensions. The values for the custom 'words' are chosen by me, and if you disagree feel free to propose changes, inclusion of new 'words' or entire new dictionaries for channels.

You must have a Twitch account and request an authentication token from https://twitchapps.com/tmi/ before use.

Enter your Twitch nickname and token into the "twitch_api_cred.txt" file.

Run python chat_sentiment_analysis -vod [video_id] to analyse a VOD where video_id is the number in the URL of the twitch VOD.

Run python chat_sentiment_analysis -live [channel_name] to analyse a live chat in real time where channel_name is the name of the channel's chat.

Use -m to specify a specific minute interval for the bars, default is 5.

Use -d to enter debugging mode, where unrecognized (single word, neutral) chats are displayed. Useful for finding things to add to the vader dictionaries.

* Live chat plotting window not responding, but still works fine. 
* Get channel name through Twitch API using channel_id instead of scraping it, since scraping seems volatile (and slower)
* Expand VADER dictionaries