Exploring extremism on YouTube collaboration methods to ensure open and reproducible data research.
For each working group please create your own folder on the master branch where to place your scripts
- Please use a descriptive folder name e.g. data gathering, topic modeling, etc.
- Please add a readme to the folder where you describe:
- Your team members and contact information
- Your approach; specifying which data you used as input, how you processed it (with link to script adn what frameworks are used), what your output data is and a link to where it is saved in the GDrive.
- You may create a branch for your subfolder to work on, so you do not have to synchornize all folders and scripts.
-
Data gathering
- Video data (metadata, transcript, comments)
- Channel data (videos with their data)
- User data (subscriptions, videos likes)
-
Topic mapping (can we identify what topics are prevalent in entities?)
- Channel topic mapping
- User topic mapping
- Video topic mapping
-
Clustering (can we identify entities that cluster around certain features?)
- Channel clustering
- User clustering
- Video clustering
-
Classification (can we classify entities to show similarities with certain topics?)
- Channel classification
- User classification
- Video classification
-
Network analysis (can we identify cliques of channels or users isolated around certain topics?)
- Channel network
- User network
- Let's standardise how and where we save data. To ensure plaftorm and technology interoperability the proposal is to save all results as CSV files with the first row having column IDs.
- In your readme you can add additional information describing the columns.
- Temporary solutions is to dave your results as CSV on GDrive. Please create direct links from GitHub readme and scripts to the file to save browsing time.
- We are exploring http://data.world as next solution; it allows you to connect through API to read and write results there.