[Developers] Ambiguities in /film domain

[Christopher R. Maden]

Dave Boyce wrote:
> I'm working on a movies viewer, and I'm finding a number of data issues.
> 
> 1. A person can be represented by more than one object.
> 
> 2. Two people with the same name result in data that is scattered, and  
> there's
> a lot of ambiguity.
> 
> I appreciate the way in which this situation has arisen and also that  
> there isn't
> an immediate or easy fix, but this is a problem. I guess this isn't  
> news, so
> does Freebase have any plans to rectify these kinds of issues, or is  
> it just
> going to be left to the community to fix them as they're encountered?

The community *is* the plan.

Some things can be detected automatically; for instance, two different 
directors with the same name on one film is probably a sign of a data 
problem.  There is an internal tracking issue to look for problems like 
that.  However, most of these problems will rely on someone like you to 
notice them and flag them for correction.

[Most of these sort of problems come into being from our Wikipedia 
import and analysis.  Someone adds a simple wikilink like [[Paul 
Verhoeven]] which selects the wrong one; someone else corrects it, but 
we’ve already captured the bad association.  We aren’t yet good at 
*correcting* errors like that, so we end up with two identically-named 
directors on the film.]

So please do keep flagging these as you find them.  Thanks.

~Chris
-- 
Christopher R. Maden
Data Architect
Freebase.com: <URL: http://www.freebase.com/ >
Metaweb Technologes, Inc. <URL: http://www.metaweb.com/ >