I’m currently in a sociology class about social networks, and my group is interested in mapping out Twitch communities. We were inspired by a project made by Kiran Gershenfeld, who took the top 100 streams and their viewers, found overlapping viewers from these streams, and created a map visualizing a snapshot of Twitch viewership. Here’s the article where Kiran describes his project:
Insights From Visualizing Public Data on Twitch
Ideally, we want to do what Kiran did, except conduct network analysis on current-day Twitch communities. However, we aren’t sure how to interact with the API in order to get data we’d need to do so. We’re novices to using API’s and Python in general, so any insight/advice on where we should start, or how we can create the dataset would be much appreciated. Thanks!
Please note that what you are planning to do is not permitted according to my understanding of the Third Party Developer Agreement and will require previous written permission from Twitch.
If you have such permission, your best bet to find out who is watching who will be to connect to chat and see who is chatting - this number however will be heavily skewed as “being in chat”, “chatting” and “watching the stream” are very, VERY distinctive things and do not necessarily correlate. The latter you cannot get from Twitch, the second is gatherable if you are listening to chat when it happens and the first one is not available from officially supported sources.
Thus, as much as I want to be welcoming and support new people working with the Twitch APIs, I cannot recommend this specific type of interaction.
(I am not a Lawyer and this is not Legal Advice.)