Also i did some digging, and despite a paywall and some articles without links at all, their publications raise more questions than they answer about the actual trust/governance model and privacy implications (see my other comments quoting some bits).
Moreover, the sentiment/subjectivity analysis via ML leaves me dubious. I have a hard time believing AI can understand human discourse and/or help solve human problems, see also James Mickens USENIX talk "Why do keynote speakers keep suggesting we can improve security?"
Finally, let me put on my paranoid hat: if i were to build a large-scale surveillance network to expose the social graph and flow of information (think XKEYSCORE) for decentralized social media, it would look a lot like EUNOMIA! On the contrary, if i were to build a reputation protocole for decentralized social media, it would be much simpler than the EUNOMIA project, and would probably be based on some kind of Web of Trust / Fog of Trust (since it's aimed at classifying/auditing public data, where the social graph is already, at least partially, exposed).
> How are these votes stored?
From what I can tell, this is literally a research project attempting to answer these questions. Look at the results tab, then under 'Demonstrators, pilots, prototypes' to see what they have come up with so far.