We present DocRT, a distributed gossip-based platform for content sharing enabling scientific document search and recommendation. DocRT combines advantages from centralized and P2P systems.
Each time a scientist shares a new article, the content, the title and tags are automatically extracted from it. Title and tags can be modified. The scientist's profile is updated based on the tagged. Search results take into account the user's profile.
A site regroups 1 to m users. Each site is part of large scale sharing network of DocRT platforms. The keywords queries are routed among diversified and chosen sites to retrieve relevant documents.
Each site periodically gossips to establish both the overlay among virtual nodes (users) and to compute statistics about the global corpus, thus, enabling a fine indexing.
The full prototype is developped in Java EE and relies on a MongoDB database.
Deploying DocRT in a network only takes minutes. A README file explains the installation and reployement procedure.