RE: [orms] Reputation Model + Comments

["Giles Hogben" <Giles.Hogben@enisa.europa.eu>]

Title: RE: [orms] Reputation Model + Comments

1. What are your thoughts on how to reflect the "strength of belief" in the reputation vote?

Just to be clear - the belief is in the assertion and this is what the vote expresses (how much you believe the assertion is true). The strength of belief is just the number of stars you give someone or whatever. In all reputation systems I know it can easily be modeled as a number between 0 and 1 - e.g. 3 stars out of 5 is 0.6 , etc...

2.What are your thoughts on how to reflect the reputation of the voter in that entity's reputation vote?

For reputation of voter, I suggest to completely separate votes from the algorithm used to calculate an aggregate score on an assertion. So you have a heap of votes, some of them about the assertion subject and some of them about the voters. All algos take that heap of votes as an input and define how to calculate the aggregate score for a given assertion from that heap of votes. Some algos will weight scores on the basis of strength of belief in certain assertions (vote scores between 0 and 1) about the voters themselves, but that info is contained in the algo definition...

Does that make any sense?


-----Original Message-----
From: Chris Daly [mailto:rcdaly@us.ibm.com]
Sent: Wed 18/06/2008 22:31
To: Giles Hogben
Cc: orms@lists.oasis-open.org
Subject: Re: [orms] Reputation Model + Comments



Regards,
Chris Daly
Technical Competency Lead for Security - IBM Software Group
tel:  301-240-2610 t/l 372
fax:  301-240-3350 t/l 372
cell: 301-524-6691
Gaithersburg, MD




"Giles Hogben" <Giles.Hogben@enisa.europa.eu>
06/18/2008 10:34 AM

To
<orms@lists.oasis-open.org>
cc

Subject
[orms] Reputation Model + Comments






OK - I'm now posting this as a member of the group

Hi All,
I've just joined this group with Observer status since I don't have a lot
of time to work on it. I work for the European Network and Information
Security Agency (ENISA) and have some experience in =
reputation systems esp security and privacy issues (a report I co-edited =
is one of your refs). I wanted to contribute a couple of things.

1. A while back I sketched (see attachment) - a language which fits a lot
of your use-cases - it's quite skeletal but I think it has some useful
ideas. Don't be put off by the vote terminology - it's about reputation,
not e-voting. This was just a personal project and not part of any larger
initiative.
2. I saw some of the discussions and I wanted to say I think it's very
important that any protocol is symettric - in the sense that you can a.
nest/recurse reputation as much as you want and b.check reputation =
reciprocally.
3. Reputation is fundamentally about agreggated beliefs on an assertion.
There should therefore be work on integrating with SAML.

Any thoughts on the attached welcome. I'd be happy to join a call and take
people through it some time (but not in the next couple of weeks probably
as I'm very busy),

Regards,

Giles Hogben
(See attached file: reputation2.pdf)
---------------------------------------------------------------------
To unsubscribe from this mail list, you must leave the OASIS TC that
generates this mail.  You may a link to this group and all your TCs in
OASIS
at:
https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php