Re: Holdings-as-Offer: wrap-up

I'm suggesting that the system and the branch would be separately identified organizations. They could be located in different places, have different contacts, etc. The clue that one is the branch of the other could boil down to that one schema:branchOf assertion connecting their identifiers.

Sent from my iPad

> On Oct 19, 2013, at 4:39 PM, "Karen Coyle" <kcoyle@kcoyle.net> wrote:
> 
> Jeff, not sure what you mean by "structured." I think your example illustrates my concern. Branch name isn't sufficient in itself. We may need to supply something else from the context of the web site.
> 
> kc
> 
>> On 10/19/13 1:26 PM, Young,Jeff (OR) wrote:
>> I don't think we should assume that branch names are structured. I checked out a book yesterday at "The Arcanaeum", which is a branch of "The College of Winterhold Library System".
>> 
>> Sent from my iPad
>> 
>>> On Oct 19, 2013, at 4:17 PM, "Karen Coyle" <kcoyle@kcoyle.net> wrote:
>>> 
>>> Jeff, I'm trying to make a more practical argument, which is that without the name of the library, the branch name isn't meaningful. I don't think we can decide what is a proper seller v. location, but if the locations aren't meaningful alone, then we need the library name.
>>> 
>>> kc
>>> 
>>>> On 10/19/13 12:59 PM, Young,Jeff (OR) wrote:
>>>> Schema.org <http://Schema.org> has a branchOf property that probably has
>>>> a role to play sorting this out.
>>>> 
>>>> I would argue that a "library system" like NYPL (in general) is more
>>>> like an administrative agency than it is an agency where books are being
>>>> circulated. The fact that the word "library" appears in the name
>>>> shouldn't be a huge problem for search engines if the data publishers
>>>> are careful.
>>>> 
>>>> Jeff
>>>> 
>>>> Sent from my iPad
>>>> 
>>>> On Oct 19, 2013, at 3:08 PM, "Dan Scott" <denials@gmail.com
>>>> <mailto:denials@gmail.com>> wrote:
>>>> 
>>>>> Also, I haven't added availableAtOrFrom to the examples on the
>>>>> Holdings proposal page because I was asking for consensus and
>>>>> providing sample examples in this thread.
>>>>> 
>>>>> I am worried about this distinction you're introducing between branch
>>>>> and library. To me, a branch is a library. The seller represents the
>>>>> current physical location of the item (if that item is physical) where
>>>>> an interested party can pick it up. I think search engines are trying
>>>>> to satisfy an immediate need, not saying "oh it's available in NYPL
>>>>> somewhere".
>>>>> 
>>>>> On Oct 19, 2013 2:24 PM, "Karen Coyle" <kcoyle@kcoyle.net
>>>>> <mailto:kcoyle@kcoyle.net>> wrote:
>>>>> 
>>>>>    Dan, I'm obviously having trouble explaining my point. Here's what
>>>>>    I would add to the example:
>>>>> 
>>>>> 
>>>>> 
>>>>>          <div itemprop="offers" itemscope
>>>>>        itemtype="http://schema.org/__Offer <http://schema.org/Offer>">
>>>>>             <meta itemprop="businessFunction"
>>>>>        content="http://purl.org/__goodrelations/v1#LeaseOut
>>>>>        <http://purl.org/goodrelations/v1#LeaseOut>">
>>>>>             <meta itemprop="seller" content="Example Branch 1</meta>
>>>>>             <div>Call number: <span
>>>>>        itemprop="inventoryIdentifier"__>876.54</span></div>
>>>>>             <div>Location: <span
>>>>>        itemprop="availableAtOrFrom">__Reference</span></div>
>>>>>             <div>Item status: <span>
>>>>>                 <link itemprop="availability"
>>>>>        href="http://schema.org/__InStoreOnly
>>>>>        <http://schema.org/InStoreOnly>">
>>>>>                 Library use only
>>>>>             </span></div>
>>>>>          </div>
>>>>> 
>>>>> 
>>>>> 
>>>>>     <div itemprop="offers" itemscope
>>>>>    itemtype="http://schema.org/__Offer <http://schema.org/Offer>">
>>>>>        <meta itemprop="businessFunction"
>>>>>    content="http://purl.org/__goodrelations/v1#LeaseOut
>>>>>    <http://purl.org/goodrelations/v1#LeaseOut>">
>>>>>        <meta itemprop="seller" content="http://sfpl.org"></__meta>
>>>>>        <div>Call number: <span
>>>>>    itemprop="inventoryIdentifier"__>876.54</span></div>
>>>>>        <div>Location: <span
>>>>>    itemprop="availableAtOrFrom">__Reference</span></div>
>>>>>        <div>Item status: <span>
>>>>>            <link itemprop="availability"
>>>>>    href="http://schema.org/__InStoreOnly
>>>>>    <http://schema.org/InStoreOnly>">
>>>>>            Library use only
>>>>>        </span></div>
>>>>>     </div>
>>>>> 
>>>>>    "Seller" becomes the library, not the location within the library,
>>>>>    and not even the branch.
>>>>> 
>>>>>    I see a useful interplay between "seller" and "availableAtOrFrom".
>>>>>    If we add "availableAtOrFrom" (which I don't see on the examples
>>>>>    on the web page for the proposal, but it makes good sense here),
>>>>>    then we have "availableAtOrFrom" for the displayed physical
>>>>>    location, whatever it is. Seller then becomes something that
>>>>>    identifies the library qua organization, and should be an
>>>>>    organization, not a location (like "Reference" or even "West
>>>>>    branch"). Seller and availableAtOrFrom might be the same, but
>>>>>    that's not a problem. For most displays, though, I think that they
>>>>>    would be different, since the holdings display doesn't usually
>>>>>    contain the library name:
>>>>> 
>>>>>    (site is Berkeley Public Library Catalog)
>>>>>    Location        Call no.        Status
>>>>>    CENTRAL LIB     876.54          Check shelf
>>>>> 
>>>>>    Imagine how many libraries have a location called "MAIN" or
>>>>>    "CENTRAL" -- that's a location, not an organization. So you need
>>>>>    to get the library organization name in there.
>>>>> 
>>>>>     <div itemprop="offers" itemscope
>>>>>    itemtype="http://schema.org/__Offer <http://schema.org/Offer>">
>>>>>        <meta itemprop="businessFunction"
>>>>>    content="http://purl.org/__goodrelations/v1#LeaseOut
>>>>>    <http://purl.org/goodrelations/v1#LeaseOut>">
>>>>>        <meta itemprop="seller" content="Berkeley Public Library"></meta>
>>>>>      <!-- or ... content="http://__berkeleypubliclibrary.org
>>>>>    <http://berkeleypubliclibrary.org>"  -->
>>>>>        <div>Call number: <span
>>>>>    itemprop="inventoryIdentifier"__>876.54</span></div>
>>>>>        <div>Location: <span itemprop="availableAtOrFrom">__CENTRAL
>>>>>    LIB</span></div>
>>>>>        <div>Item status: <span>
>>>>>            <link itemprop="availability"
>>>>>    href="http://schema.org/__inStock <http://schema.org/inStock>">
>>>>>            Check shelf
>>>>>        </span></div>
>>>>>     </div>
>>>>> 
>>>>>    Does that make sense?
>>>>> 
>>>>>    kc
>>>>> 
>>>>> 
>>>>> 
>>>>>        I can include an example to cover this use case.
>>>>> 
>>>>>            Another case is for electronic
>>>>>            materials. Library systems handle this differently, but
>>>>>            there isn't a
>>>>>            location in many cases:
>>>>> 
>>>>>            Online    Click here
>>>>> 
>>>>> 
>>>>>        Good question! What we do in Evergreen currently is essentially:
>>>>> 
>>>>>        <li property="offers" vocab="http://schema.org/" typeof="Offer">
>>>>>           <a href="URL_FOR_ELECTRONIC___MATERIAL">Available online</a>
>>>>>           <link property="availability"
>>>>>        href="http://schema.org/__OnlineOnly
>>>>>        <http://schema.org/OnlineOnly>" />
>>>>>           ...
>>>>>        </li>
>>>>> 
>>>>>        Glancing at this, I think I need to add in a property="url" to
>>>>>        the <a>
>>>>>        link there. And it certainly can have a <meta> tag for the
>>>>>        "seller"
>>>>>        property as well. I can include some documentation and an
>>>>>        example for
>>>>>        this use case, too.
>>>>> 
>>>>>            I'm thinking that there are cases in which the library
>>>>>            itself is not
>>>>>            included in the holdings statement (or anywhere else on
>>>>>            the page) because it
>>>>>            is inherent in the context of the system being searched.
>>>>>            So my question is
>>>>>            whether there is value in including information about the
>>>>>            library itself as
>>>>>            a super-location to the holdings location, or is the
>>>>>            assumption that this
>>>>>            connection will be made through, e.g., the URL of the web
>>>>>            page that has the
>>>>>            markup?
>>>>> 
>>>>> 
>>>>>        I think there is significant value to creating 1) a page per
>>>>>        library
>>>>>        for a given system that contains the physical addresses /
>>>>>        hours / etc,
>>>>>        even for single-library systems and 2) linking to that page
>>>>>        (explicitly, or implicitly via <meta>) from each of the associated
>>>>>        offers. It will enable the search engines to follow their nose
>>>>>        based
>>>>>        on our assertions, rather than having to make assumptions
>>>>>        about how
>>>>>        many libraries may inherently be represented by
>>>>>        library.example.com <http://library.example.com>.
>>>>> 
>>>>>            I think my question leads to a broader one about the use
>>>>>            case for library
>>>>>            data in schema.org <http://schema.org>. When I look at
>>>>>            product examples it is clear to me that
>>>>>            the target is the URL of the product page. Is this also
>>>>>            the assumption for
>>>>>            library data in schema.org <http://schema.org> -- that we
>>>>>            are expecting a search engine
>>>>>            retrieval of a page for a library resource, and that page
>>>>>            is the target of
>>>>>            the search? If so, then that URL is all that is needed to
>>>>>            link to the
>>>>>            library and its resource. If, however, we anticipate other
>>>>>            uses to be made
>>>>>            of the schema mark-up, such as organizing retrieved items
>>>>>            by geographical
>>>>>            location, then we need to get that information into each
>>>>>            web page. This may
>>>>>            be unrelated to the markup of holdings, but it was this
>>>>>            proposal that
>>>>>            brought it to mind.
>>>>> 
>>>>> 
>>>>>        Good question again. I see the primary use case being the search
>>>>>        engines ingesting a sitemap, crawling all of the listed pages, and
>>>>>        sorting out the items and linked offers accordingly.
>>>>> 
>>>>>        In the case of our library catalogue, I created a sitemap that
>>>>>        lists
>>>>>        each of the record detail pages, which expose metadata & holdings.
>>>>>        Unfortunately, when I generated the sitemap last year, it was
>>>>>        before I
>>>>>        had implemented holdings-as-offers; now that all of the attached
>>>>>        offers will be part of each record details page, I think a new
>>>>>        crawl
>>>>>        of those pages could provoke much more interesting results.
>>>>> 
>>>>> 
>>>>>    --
>>>>>    Karen Coyle
>>>>>    kcoyle@kcoyle.net <mailto:kcoyle@kcoyle.net> http://kcoyle.net
>>>>>    m: 1-510-435-8234 <tel:1-510-435-8234>
>>>>>    skype: kcoylenet
>>> 
>>> --
>>> Karen Coyle
>>> kcoyle@kcoyle.net http://kcoyle.net
>>> m: 1-510-435-8234
>>> skype: kcoylenet
> 
> -- 
> Karen Coyle
> kcoyle@kcoyle.net http://kcoyle.net
> m: 1-510-435-8234
> skype: kcoylenet
> 

Received on Saturday, 19 October 2013 20:45:07 UTC