Roadmaps

The core of Metacademy is the concept graph, which has structured information about relationships between different concepts and how to learn about them. If the user has something in particular they're hoping to understand, they can hopefully find what they're looking for by searching for the concept. But often our learning goals are more nebulous: we know we want to learn about a general area because it's relevant to something we've just read, or to a problem we're trying to solve, but it's not clear where to start.

This is where roadmaps come in. A roadmap is basically a document which gives the reader an overview of some topic, with links to relevant concepts and outside resources. Here are some hypothetical examples we have in mind:

• An overview of a broad research area, such as deep learning. This would include Metacademy concepts, research papers and review articles which explain the important results, and a discussion of current open questions. For each of the academic papers, it would give the set of concepts you should learn first in order to make sense of the paper.
• A hands-on tutorial geared towards an application area, such as "build a recommendation system." The tutorial would guide the reader through a project in a series of steps, with pointers to Metacademy concepts as they're needed. The first step, for instance, might be to learn about and implement vanilla linear regression. Then in the next step, you observe that it's overfitting, learn about regularization, and fix the problem.

There's nothing technologically fancy here: in our current conception, it's just a wiki page. But we're hoping that with the concept graph as a foundation, it will be easy for experts to put together roadmaps for all sorts of different audiences.