Re: a user liking a post

On 2015-08-23 10:21, Melvin Carvalho wrote:
> 2. The semantics of a post liking another post seems slightly unnatural
> to me.  I guess in this case the post is used as an indirect identifier
> for a person (or account).  Intuitively I think it may be confusing for
> software to conflate these concepts, but if it's working for some,
> that's great.  So slightly cautious regarding reuse, as I tend to have
> slightly more specific semantics (ie person oriented rather than post
> oriented)

I agree that this is not the most intuitive approach. After all, the 
natural language (or at least what's visible for humans) roughly says 
"Melvin likes this", and not "Something in which Melvin is the author of 
indicates that it likes something else". (Aside: we'll come back to 
temporality).

AS2's as:Like and IWC's like-of essentially follow the same approach, 
i.e., qualified relations. I don't think the particular location of the 
"liking" makes a difference other than how they may be 
discovered/queried for.

The machine representations look awkward because we are not actually 
structuring the human language. It is just an alternative perspective on 
what's happening.

Having said that, when as:Like is accompanied with the actor (a 
person/agent..) and an object (e.g., photo) it gives us the same 
information. The small (?) difference with IWC's approach is that, it is 
specifically about an object (i.e., an entry) being a like-of another 
object. (Aside: AS activities are objects as well.. so, it actually 
boils down to both approaches being qualified relations).

Coming back to temporality:

IMO, this is where things get interesting. Activity centric view with 
qualified relations we can be more explicit or extend with the time 
dimension e.g., "Tantek started to like 'The Matrix' on 1999-03-31".

Consequently, we don't actually have to come up with past/current/future 
terms for all activities. We simply use the core concept and attach time 
information.

This activity view is quite flexible because we can also capture the 
evolution of the objects e.g., wiki-like changes, repository history, 
chat history, with no change to the underlying model, and they are still 
a form of social activities. This also implies that something is 
influenced by another thing. Enter data provenance.


PS: Ran into this vocab recently: http://purl.org/net/soron "complements 
FOAF and RELATIONSHIP ontologies", where subject is a person. IMO, there 
is no need for these type of vocabularies as they simply don't / can't / 
shouldn't try to capture all human interactions.

-Sarven
http://csarven.ca/#i

Received on Sunday, 23 August 2015 09:45:55 UTC