We want Mapedia to be an open project. To us, this means using crowd-sourcing to build a platform promoting free education, being transparent on all related issues, and remaining independent and mission-driven. Additionally, we tend to approach things from a pragmatic standpoint. This in particular motivates our choice of focusing first on domains affiliated with tech and on indexing and curating external content in a way that benefits the learner right from the start.
Mapedia is an open, collaborative, and community-based project. We aim to leverage collective intelligence to crowdsource ideas, receive quality feedback from diverse perspectives, and find help to solve some of the most challenging problems online learning faces. We have a Discord and you can find a guide for contributing to the project here.
The code is open source and the content internal to the platform (user-generated) will be free. We also give priority to free external content, unless there is no good unpaid option available or a paid resource is the reference to learn a specific concept and just can't be omitted. We realize that, sometimes, spending a few bucks on a great book is the best way to go.
We believe transparency to be central to community-based projects. In particular, this means explaining our decisions and communicating regularly about the state of the project. We aspire to be transparent about the inner logic of the recommendation algorithm and provide users with control over it.
Learning from the failures of similar endeavors, we intend to remain as independent as possible from public institutions and develop this platform with a bias for action.
For the project to take off, we need to create momentum and build the community. Any constructive feedback is very valuable at this stage and it is a great starting point if you are interested in getting involved.
We are currently self-funded but plan to apply for social entrepreneurship programs and to later organize a crowdfunding campaign in order to scale.
We have several options for monetization in the long run and prefer not to focus on this aspect at the moment, acquiring a user base and building a community being the priority.
In order to have user-generated content, Mapedia first needs a strong user base. As a first step in this direction and in order to offer an experience that is beneficial even before reaching critical mass, we will first support learners by indexing free external content and serving it in a valuable way, through recommendation engines and curated learning paths. The next milestones will be to gather feedback on resource quality, add social features, and drive contributions towards user-generated content. We are aiming for exponential growth powered by network effects.
Our initial focus will mainly be around tech domains. The reasons behind this choice are that the founders are more familiar with these topics and experts in some, that demand is strong, and that mastering a lot of these skills is necessary to build Mapedia. Besides, this approach puts us in the learner's shoes and makes us users of our own product.
As we want to avoid the pitfall of designing a model of the knowledge space that is too specific to tech and not abstract enough, we will quickly support diverse other domains such as behavioral sciences, biology, languages, or philosophy.
The first release (August 2020) consisted of a recommendation feed of free, external learning resources for each supported domain. The recommendations are based on a mapping of dependencies between concepts and the learner's prior knowledge.
Since then, we have implemented the possibility to create and share curated learning paths (November 2020).
We are currently developping features to manage learning goals, along with prerequisites and outcomes of learning materials.
The next steps include improving the recommendations to make them more valuable, expanding on learning path functionalities, and indexing more resources.
We are not idealists and believe that, in order to succeed in such an ambitious enterprise, it is mandatory to always stay pragmatic:
We are modeling the knowledge space as graph, using a graph database (namely Neo4j). Graphs are particularly well-adapted for this purpose, and provide relatively easy yet flexible recommendations functionalities, with little need for user data.
We plan to support collaborative work with automation tools, for instance, to figure out which concepts constitute a domain or to find and to analyze learning resources. Those tools would mainly involve web crawlers and machine learning (e.g. Natural Language Processing). If you have experience in that regard, we are very keen to hear from you.
We're particularly thrilled to have been granted access to OpenAI's GPT-3 API and are very excited about what we have been able to test so far.