[Data-modeling] Are Influence Nodes necessarily persons?

Gordon Mackenzie gordon at metaweb.com
Fri Oct 31 08:08:55 UTC 2008


I always took peer as a necessary part of the influence node...There are some relationships in influence where source ---> recipient  is not valid, but influence was present, usually bi-directionally.

----- Original Message -----
From: "Alexander Marks" <al at metaweb.com>
To: "Freebase data modeling mailing list" <data-modeling at freebase.com>
Sent: Thursday, October 30, 2008 5:54:17 PM GMT -08:00 US/Canada Pacific
Subject: Re: [Data-modeling] Are Influence Nodes necessarily persons?

I think the peer property should be deprecated because it's a denormalization that makes every query to the model a painful two step process; you can't ask "who influenced X?" without also asking "who is a peer of X?".

I would support expanding influence node to every type, since it's easy to add a /people/person type constraint, but I wouldn't support adding any more types to the model.

Al

----- Original Message -----
From: "Jeff Prucher" <jeff at metaweb.com>
To: "Freebase data modeling mailing list" <data-modeling at freebase.com>
Sent: Thursday, October 30, 2008 5:37:56 PM GMT -08:00 US/Canada Pacific
Subject: Re: [Data-modeling] Are Influence Nodes necessarily persons?

I wasn't actually proposing making Peer a type, so you can't second it!  I
was raising it as an objection to deleting Influence Node because we'd be
replacing one type with three. 

> -----Original Message-----
> From: data-modeling-bounces at freebase.com 
> [mailto:data-modeling-bounces at freebase.com] On Behalf Of Faye Harris
> Sent: Thursday, October 30, 2008 5:31 PM
> To: Freebase data modeling mailing list
> Subject: Re: [Data-modeling] Are Influence Nodes necessarily persons?
> 
> Ah yes, "Peer" -- I'd second it as a type. I knew I was 
> forgetting something.
> 
> And yes, breaking James's application on who influenced whom 
> would be a concern. It's unfortunate. But I do feel that this 
> would constitute a necessary move in the right direction. 
> Data on influence between two persons would be an easily 
> query-able subset of data on influence between any two topics 
> of arbitrary type.
> 
> The resulting ability in the new schema to model influence 
> between any two instances would be a clear gain.
> 
> -- Faye
> 
> 
> Jeff Prucher wrote:
> > I think the phylogeny pattern fits this type of data better than a 
> > two-type system. Because influence is a chain (X influence Y who 
> > influenced Z), a phylogeny is a lot cleaner; otherwise Y 
> needs to be 
> > both an influencee and an influencer, which seems unnecessarily 
> > complicated.  Influence node also has the Peers property, 
> which would 
> > presumably have to be handled by another type (Peer?), if 
> we get rid of the Influence Node type.
> >
> > As regards person vs. other influencable thing, I don't 
> have a strong 
> > opinion, but I do want to note that there are applications that use 
> > this model which may be affected by changing the usage.
> >
> > Jeff
> >
> >   
> >> -----Original Message-----
> >> From: data-modeling-bounces at freebase.com
> >> [mailto:data-modeling-bounces at freebase.com] On Behalf Of 
> Faye Harris
> >> Sent: Thursday, October 30, 2008 4:59 PM
> >> To: Freebase data modeling mailing list
> >> Subject: Re: [Data-modeling] Are Influence Nodes 
> necessarily persons?
> >>
> >> I've actually given this some thought recently, hence the 
> relatively 
> >> quick response. :)
> >>
> >> As data models go, "Influence Node" is sort of an odd ball. 
> >> It doesn't tell you immediately if an instance is an 
> "influencer" or 
> >> "influencee"
> >> (pardon the coinage), just that it has data related to influence. 
> >> Normally, types in Freebase model an "Is-A" relationship. San 
> >> Francisco "Is-A" location, and we apply the "Location" type to the 
> >> topic for San Francisco to represent that relationship.
> >>
> >> So a comparable schema for influence would ideally have separate 
> >> "Influencer" and "Influencee" (or "Influenced", or whatever name) 
> >> types, applied to topics where... well, applicable. As a 
> generalized 
> >> concept the included type of "Person" is rather unnecessary. 
> >> Influence may be between two things of the same type, or 
> any type, or 
> >> between things of different types. A film can influence an 
> author, a 
> >> philosophy can influence an art movement, etc.
> >>
> >> If asked between 1) remodeling the current "Influence Node" 
> >> type to two separate types without "Person" as an included 
> type, and 
> >> 2) creating a new pair of types to duplicate the same influence 
> >> relationship without "Person" as an included type, I'd choose the 
> >> former. It'd take a little reshuffling, but I think it'd be a lot 
> >> less confusing in the long run, than having two sets of similar 
> >> types, one a subset of the other.
> >>
> >> -- Faye
> >>
> >>
> >> Tadhg O'Higgins wrote:
> >>     
> >>> Would Influence Node be more useful/interesting if it 
> didn't assume 
> >>> personhood for its instances? Books, Films, Games, and
> >>>       
> >> Musical Groups
> >>     
> >>> are all examples of things that seem like Influence Nodes
> >>>       
> >> but are not
> >>     
> >>> persons.
> >>>
> >>> We should consider removing the /people/person included type from 
> >>> Influence Node for that reason. The alternatives I can see
> >>>       
> >> are to a)
> >>     
> >>> create an "Influence <thing>" in each domain that might
> >>>       
> >> support one,
> >>     
> >>> or
> >>> b) create an "Influence Thing" type that's just like 
> Influence Node 
> >>> but doesn't include personhood. Neither of those seem as good a 
> >>> solution as widening the scope of Influence Node.
> >>>
> >>> Thoughts?
> >>>
> >>> Tadhg
> >>> _______________________________________________
> >>> Data-modeling mailing list
> >>> Data-modeling at freebase.com
> >>> http://lists.freebase.com/mailman/listinfo/data-modeling
> >>>
> >>>   
> >>>       
> >> _______________________________________________
> >> Data-modeling mailing list
> >> Data-modeling at freebase.com
> >> http://lists.freebase.com/mailman/listinfo/data-modeling
> >>
> >>     
> >
> > _______________________________________________
> > Data-modeling mailing list
> > Data-modeling at freebase.com
> > http://lists.freebase.com/mailman/listinfo/data-modeling
> >
> >   
> 
> _______________________________________________
> Data-modeling mailing list
> Data-modeling at freebase.com
> http://lists.freebase.com/mailman/listinfo/data-modeling
> 

_______________________________________________
Data-modeling mailing list
Data-modeling at freebase.com
http://lists.freebase.com/mailman/listinfo/data-modeling
_______________________________________________
Data-modeling mailing list
Data-modeling at freebase.com
http://lists.freebase.com/mailman/listinfo/data-modeling


More information about the Data-modeling mailing list