Dataset

We have collected a sample dataset from a previous existing project. The datasets presented below are for academic use only. The crawled data were continually collected from Twitch every five minutes in a one-month period (Feb. 1st-28th, 2015). Through the official APIs, a multi-thread crawler obtained information from each broadcaster and the official system dashboard. The crawler does not need Twitch's API client-ID and avoids the limitation for the maximum number of objects to return in each request. These are the contents of the dataset used in our project:

Name Description
Current Views An integer number of current viewers.
Game Name A string of game name.
Follower Number An integer number of the followers.
Broadcaster Language A string of broadcaster's language.

The dataset spans across 20000+ txt files, weighting over 15GiB.