Print

Print


To provide an example of this, I suggest considering 
https://www.freebase.com/m/0c43g?links= Freebase is a little like 
Wikidata, but with an emphasis on importing and cross matching datasets 
with semi-automated matching and less in the way of editorial oversight 
('lack of provenance' in library-speak).

Most of those 591 links look fine, but some of them I'm struggling to 
understand, let alone figure out work out whether they're relevant to a 
library catalog or how to present to an end-user.

However, as a fail-over for when wikidata doesn't have what a cataloguer 
needs, Freebase is entirely acceptable. It may also be suitable for 
specialist tools that avoid taking all assertions at face value.

cheers
stuart

On 07/11/2014 06:33 AM, [log in to unmask] wrote:
> It is useful and convenient, and it won't serve all purposes-- it's good to have, it's just not enough.
>
> That second principle stands entire in a world in which provenance information is separate from DNS. Let's say you publish an HTTP URI identifying an author interesting to me, and I make some assertions using your URI. Then someone consuming my data can "travel through" that URI to find out _something_ more about that author, which is to say, to discover more assertions involving that author (presumably with a subject of that author). So far so good, and that's as far as the second principle takes us. Provenance becomes interesting when my hypothetical patron asks, "Wait, how much can I trust this information about my author? Of what value are these assertions published by Jeff Young to _me_?"
>
> Now, we're no longer in the realm of questions that can be answered by the DNS.
>
> I wouldn't propose that we somehow "outlaw" letting assumptions about provenance rely on DNS, and we couldn't if I was odd enough to demand it. I do propose that we frequently need more than that, and that we should be thinking about how to do it. That doesn't seem too controversial to me. {grin}
>
> ---
> A. Soroka
> The University of Virginia Library
>
> On Jul 10, 2014, at 2:21 PM, "Young,Jeff (OR)" <[log in to unmask]> wrote:
>
>> It depends on how strictly you interpret "provenance" and the need/willingness to carry it's burden. If people are overly concerned about DNS, then what's the point of TimBL's 2nd principle of Linked Data?
>>
>> 2. Use HTTP URIs so that people can look up those names.
>>
>> This isn't a heavy-weight form of provenance and won't serve all purposes, but it's useful and convenient.
>>
>> Jeff
>>
>>
>>> -----Original Message-----
>>> From: Bibliographic Framework Transition Initiative Forum
>>> [mailto:[log in to unmask]] On Behalf Of [log in to unmask]
>>> Sent: Thursday, July 10, 2014 2:04 PM
>>> To: [log in to unmask]
>>> Subject: Re: [BIBFRAME] Bibframe and Linked Data (Authorities)
>>>
>>> No, it does not!
>>>
>>> We need to distinguish carefully between the URI, which is a pure
>>> identifier, and the URI-that-happens-to-be-an-URL, which is also a
>>> pointer or link. Much of the power of linked data comes precisely from
>>> combining those roles, but that needs to be a conscious decision and
>>> not a matter of faith. {grin}
>>>
>>> Let's say that a very small institution publishes a number of URIs like
>>> "http://www.littlelibrary.org/authorities/4535". Then Little Library
>>> disappears as an organization, and its domain is purchased by someone
>>> else. It becomes instantly possible for that someone else to publish
>>> anything at all into that namespace and, if we base provenance on the
>>> DNS, we have no way to distinguish these groups of identifiers. You
>>> might say that VIAF is unlikely to disappear tomorrow, and that's true,
>>> but the point is that relying on domain name registrars to manage the
>>> provenance information of our metadata would be an accident waiting to
>>> happen.
>>>
>>> ---
>>> A. Soroka
>>> The University of Virginia Library
>>>
>>> On Jul 10, 2014, at 1:40 PM, "Smith-Yoshimura,Karen"
>>> <[log in to unmask]> wrote:
>>>
>>>> Kevin -
>>>>
>>>> Re:
>>>>> There's been lots of talk about provenance and the like in a global
>>> graph of data, but I feel most of those discussions rely on fairly
>>> technical mechanisms, the complexity of >which outweigh the simplicity
>>> of minting one's own URI.  (Also, the provenance statements will need
>>> their own URIs!)
>>>>
>>>> Doesn't  http://viaf.org/viaf/54202464 show the provenance is VIAF?
>>> What's complex about this?
>>>>
>>>> Karen S-Y
>>>>
>>>>
>>>>
>