Embedding for supervised and unsupervised tasks
Social media represents, nowadays, one of the most interesting and rich sources of information. Every day, thousands of new connections arise, new users join communities, and billions of posts are shared. Graphs mathematically represent all those interactions, helping to make order of all such spontaneous and unstructured traffic.
When dealing with social graphs, there are many interesting problems that can be addressed using machine learning. Under the correct settings, it is possible to extract useful insights from this huge amount of data, for improving your marketing strategy, identifying users with dangerous behaviors (for example, terrorist networks), and predicting the likelihood that a user will read your new post.
Specifically, link prediction is one of the most interesting and important research topics in this field. Depending on what a connection in your social graph represents, by predicting future edges, you will be...
