We propose a framework underlying decentralized social media called SMART, capable of finding relevant interest communities without violating users’ privacy and anonymity. Its objective is to improve trust and eliminate malicious actors in participatory exchanges and collaborative decision making.
The five folders *data-hub*, *message-broker*, *rest-gateway*, *transaction-hub-in* and *transaction-hub-out* contain the microservices for the five architecture layers respectively.
The *modules* folder contains various Python modules used by multiple microservices to improve code-reuse. When building the Docker images this folder is copied into the app path.
The *tools* folder contains supportive scripts used during development and must not be used by any microservice.
The scripts *build.py* and *deploy.py* are used to create Docker images and deploy these images in Kubernetes respectively.
The *images* folder holds the images used in this document.
The semantic linking microservice helps to explicitly state implicit connections between users by analyzing extensive and repeated interactions. The explicit information about links will then be the basis for all further reasoning, classification and intelligence.
This microservice receives a message from the trace retrieval microservice when new traces are available. It then loads all traces by calling the trace retrieval's REST interface and processes these.
The trace retrieval microservice is used as an interface to the blockchain component (TIC). It receives and preprocesses all available traces.
The RESTful API gateway sends traces as messages to the trace retrieval microservice. These traces are then stored in a MongoDB document database and the semantic linking microservice is notified. This microservice provides access to all stored traces via REST.
