Reputation

From IIW
Revision as of 15:51, 9 May 2014 by Ebgross (talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Session Topic: Open Reputation Framework

Wednesday 1G http://iiw.idcommons.net/Main_Page Convener: Dave Sanford

Notes-taker(s): Dave Sanford

Discussion notes, key understandings, outstanding questions, observations, and, if appropriate to this discussion: action items, next steps:

Dave started by giving an architectural overview of a proposed Open Reputation system that would allow individuals to assert reputations to individuals, their knowledge and possibly lots of other things (e.g. products, etc.). Mostly this is not self-assertion of their own reputation, while that should probably be allowed - it has very little weight outside itself.

Dave also suggested that to be decentralized this should be build on top of a decentralized consensus algorithms like block chain or ripple.

The model includes an individual's ability to define their preferences for their own use in curating and weighting the value of information sources, etc. so that they can filter information coming in. By feeding these weighted preferences to:

Aggregate reputation modes - which use various weighting algorithms (pagerank?, Bayesian) to create weightings of reputations which are available to individuals.

Individuals and reputation nodes will have reputations that are created about the quality of the reputations that they produce, which change over time.

There were various discussions about how reputation information is defined and communicated - that included discussions about comments and context. This led to the discussion of information being communicated via graphs and XDI.

? asserted that this becomes communicated like X has a reputation for Y among Z.

Lots more discussion - common protocol is clearly required. Is a common reputation algorithm required for individual and/or aggregate reputation nodes?


Extra Notes

  • is a complex space
  • concrete use cases for killer app needed
  • Reputation score is original number
  • Lots of paper to read