W3C home > Mailing lists > Public > public-vocabs@w3.org > September 2013

Re: Semantically marking up a "checklist" or process

From: Justin Boyan <jaboyan@google.com>
Date: Tue, 10 Sep 2013 10:48:16 -0400
Message-ID: <CABJSzUvaL3UKOPh_2JPHwN3pjBBRLspMD9XdYq660aPxH3wLRQ@mail.gmail.com>
To: Martin Hepp <mfhepp@gmail.com>
Cc: Dan Brickley <danbri@google.com>, Guha Guha <guha@google.com>, Jarno van Driel <jarno@quantumspork.nl>, TallyFy <hello@tallyfy.com>, Jason Douglas <jasondouglas@google.com>, PublicVocabs <public-vocabs@w3.org>, Vicki Tardif Holland <vtardif@google.com>, Sam Goto <goto@google.com>
Martin,

Wouldn't the ListItem also need a second property, call it "item", with a
range of Thing? Otherwise how would we mark up a list of Restaurants, a
list of Universities, etc.?

Justin


On Tue, Sep 10, 2013 at 9:03 AM, Martin Hepp <mfhepp@gmail.com> wrote:

> Hi Dan, Guha:
>
> Attached, please find the proposal in the RDFa format necessary for
> inclusion in schema.org.
>
> Martin
>
>
>
>
>
> <!-- ========================== -->
> <!-- CHANGES TO EXISTING ELEMENTS -->
> <!-- ========================== -->
>
> <!-- New range
> <div typeof="rdf:Property" about="http://schema.org/itemListElement">
>         <span>Domain: <a property="http://schema.org/domain" href="
> http://schema.org/ItemList">ItemList</a></span>
>         <span>Range: <a property="http://schema.org/range" href="
> http://schema.org/ListItem">ListItem</a></span>
>         <span>Range: <a property="http://schema.org/range" href="
> http://schema.org/Text">Text</a></span>
>         ItemList
> </div>
>
> <!-- ========================== -->
> <!-- ADDITIONAL ELEMENTS -->
> <!-- ========================== -->
>
> <div typeof="rdfs:Class" about="http://schema.org/ListItem">
>         <span class="h" property="rdfs:label">ListItem</span>
>         <span property="rdfs:comment">An list item, e.g. a step in a
> checklist or how-to description.</span>
>         <span>Subclass of: <a property="rdfs:subClassOf" href="
> http://schema.org/StructuredValue">StructuredValue</a></span>
> </div>
>
> <div typeof="rdf:Property" about="http://schema.org/itemPosition">
>         <span class="h" property="rdfs:label">itemPosition</span>
>         <span property="rdfs:comment">The position of the item in an
> ordered list (1 = first, 2 = second, ...).</span>
>         <span>Domain: <a property="http://schema.org/domain" href="
> http://schema.org/ListItem">ListItem</a></span>
>         <span>Range: <a property="http://schema.org/range" href="
> http://schema.org/Number">Number</a></span>
> </div>
>
>
> On Sep 10, 2013, at 3:01 PM, Martin Hepp wrote:
>
> > Hi Jarno:
> >
> >>> Now if the expected value of an itemListElement could also be a Thing,
> wouldn't both the order (of the output array) and the Things it's about be
> preserved?
> >
> > In fact, that includes the interesting question whether the expected
> range for itemListElement could be broadened to schema:Thing.
> > From the top of my head, I would oppose that, for the simple reason that
> the property itemPosition would then have to be added to Thing, which is
> confusing.
> > Second, if you want to type the ListItem further, you could simply use a
> secondary type via basic RDFa patterns or the additionalType property.
> >
> > As for the order:
> > Implicitly, the order of the elements from the HTML tree would be
> accessible. But at least in RDFa syntax that is not preserved when the data
> is extracted.
> > Also, it is possible that the ordering in the list differs from the
> intended conceptual ordering.
> >
> > So again, I think that with as little as one new type, one range change,
> and one additional property we could get this issue done.
> >
> > Martin
> >
> >
> > On Sep 10, 2013, at 2:18 PM, Jarno van Driel wrote:
> >
> >> Correction: I should have mentioned:
> http://lists.w3.org/Archives/Public/public-vocabs/2013Jun/0042.html The
> other example contain formatting errors.
> >>
> >>
> >> On Tue, Sep 10, 2013 at 2:10 PM, Jarno van Driel <jarno@quantumspork.nl>
> wrote:
> >>>
> >>> One of the problems I tried to raise/get answered (
> http://lists.w3.org/Archives/Public/public-vocabs/2013Jun/0043.html)
> about thehttp://schema.org/ItemList itemListElement property is that it's
> expected value is text. So if you mark up a top10 list of Things you loose
> the linkage between the ItemList and the Things it's about.
> >>>
> >>>
> >>> Now if the expected value of an itemListElement could also be a Thing,
> wouldn't both the order (of the output array) and the Things it's about be
> preserved?
> >>>
> >>>
> >>> On Tue, Sep 10, 2013 at 1:51 PM, Justin Boyan <jaboyan@google.com>
> wrote:
> >>>>
> >>>> I support Martin's suggestion. This would also better model the
> common structure of "top 10 lists", such as these:
> >>>>
> >>>> http://www.zagat.com/l/boston/great-restaurants-for-ribs-in-boston
> >>>>
> http://blogs.sfweekly.com/foodie/2012/01/san_franciscos_top_10_burritos.php
> >>>>
> http://guestofaguest.com/new-york/nightlife/downtown-nyc-happy-hour-10-bars-to-check-out-after-work-today
> >>>>
> >>>> Note that in the last 2 of these 3 cases, the list is separated over
> multiple web pages, which makes it crucial to model the position number
> explicitly rather than trying to infer it from the container.
> >>>>
> >>>> I think ListItem should not be a subtype of WebPageElement - that's
> just confounding two things and adding a bunch of needless subproperties.
> It can live under schema.org/StructuredValue with other similar types.
> >>>>
> >>>> I would suggest that itemPosition be 1-based, rather than 0-based,
> since that is by far the predominant usage for all the use cases discussed
> above.
> >>>>
> >>>> Justin
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>
> >>>> On Tue, Sep 10, 2013 at 2:54 AM, Martin Hepp <
> martin.hepp@ebusiness-unibw.org> wrote:
> >>>>>
> >>>>> Hi Amit,
> >>>>> If the goal is to merely capture the elements of a checklist as a
> list structure, then
> >>>>>
> >>>>>    http://schema.org/ItemList
> >>>>>
> >>>>> should IMO provide all that is needed.
> >>>>>
> >>>>> In RDFa or other RDF syntaxes, this of course means loosing the
> order of the items, as Vicki Tardif already pointed out.
> >>>>>
> >>>>> A simple solution would be to
> >>>>>
> >>>>> 1. define a type ListItem with an additional property
> >>>>>
> >>>>> itemPosition Number The position of the item in an ordered list 0 =
> first, 1 = second, ...
> >>>>>
> >>>>> We could also reuse
> >>>>>
> >>>>>        http://purl.org/goodrelations/v1#displayPosition
> >>>>>
> >>>>> for that; it serves a similar purpose.
> >>>>>
> >>>>> 2. expand the range of the itemListElement from Text to Text or
> ListItem
> >>>>>
> >>>>> That should do the trick. At least I guess you could immediately
> mark up all of the example pages you listed.
> >>>>>
> >>>>> By the way, we should discuss whether ItemList should rather become
> a subtype of http://schema.org/WebPageElement, since we have Table there,
> so we may also want to have List there. A counter argument is that while
> Table is a significant Web page element type, List is a more generic data
> structure and not constrained to Web pages. (But then again, some tables
> outside of HTML markup, e.g. in JSON-LD or CSV, are also not
> WebPageElements in the strict sense).
> >>>>>
> >>>>> Martin
> >>>>>
> >>>>>
> >>>>>
> >>>>> On Sep 9, 2013, at 5:44 PM, TallyFy wrote:
> >>>>>
> >>>>>> Some examples  ...
> >>>>>>
> >>>>>> Web:
> >>>>>>
> http://www.realsimple.com/home-organizing/cleaning/fall-cleaning-checklist-00000000000928/index.html
> >>>>>> http://www.wikihow.com/Main-Page
> >>>>>>
> http://www.realsimple.com/weddings/dress-attire/wedding-gown-shopping-checklist-00000000000200/index.html
> >>>>>> http://terrymorris.net/bestpractices/
> >>>>>>
> >>>>>> Gov:
> >>>>>> https://www.gov.uk/foreign-travel-checklist
> >>>>>>
> https://www.gov.uk/government/publications/hurricane-supply-checklist (in
> a pdf)
> >>>>>> http://web.nvd.nist.gov/view/ncp/repository (a simpler version
> would be great!)
> >>>>>>
> >>>>>> Health:
> >>>>>>
> http://www.nhs.uk/Livewell/Healthyhearts/Pages/Arrhythmiachecklist.aspx
> >>>>>>
> >>>>>> I proposed this initiative just to wrap steps in a checklist. The
> capture of content from each step or conditional stuff is out of range and
> is a user interaction. There's many examples in the book "The Checklist
> Manifesto" by Dr. Atul Gawande:
> >>>>>> http://gawande.com/the-checklist-manifesto
> >>>>>>
> >>>>>> When Tallyfy launches in a few months, we will have some too.
> >>>>>>
> >>>>>> thanks
> >>>>>> Amit
> >>>>>> On Monday, 9 September 2013 at 15:39, Martin Hepp wrote:
> >>>>>>
> >>>>>>> Hi Jason:
> >>>>>>>> Process modeling is a rat hole and way out of scope, IMO
> >>>>>>>
> >>>>>>> I fully agree ;-)
> >>>>>>>
> >>>>>>> But even if you decide to add a very simple mechanism for exposing
> structured "step-by-step" info, I think that both
> >>>>>>> a) explicit control flows (step x follows step x) and
> >>>>>>> b( patterns for declarative approaches should be added (like
> "dependsOn" and "consequence" or"nextStep").
> >>>>>>>
> >>>>>>> Is the proposal under discussion here driven by actual use cases?
> If such, it would be good to have a couple of sites at hand that currently
> expose such checklist or process information.
> >>>>>>>
> >>>>>>> Martin
> >>>>>>>
> >>>>>>>
> >>>>>>> On Sep 9, 2013, at 4:33 PM, Jason Douglas wrote:
> >>>>>>>
> >>>>>>>> Yipes. I thought this thread was just about understanding "howto"
> content pages in a structured way. Process modeling is a rat hole and way
> out of scope, IMO.
> >>>>>>>>
> >>>>>>>>
> >>>>>>>> On Mon, Sep 9, 2013 at 4:17 AM, Martin Hepp <
> martin.hepp@ebusiness-unibw.org> wrote:
> >>>>>>>>
> >>>>>>>> On Sep 9, 2013, at 1:00 PM, Tallyfy wrote:
> >>>>>>>>
> >>>>>>>>> Are Wil and Jan members of this list?
> >>>>>>>> I don't know, but I don't think so.
> >>>>>>>>
> >>>>>>>>> Without prejudice to some work here that may result in a simple
> and web-friendly spec, I think some organisation to reach the goal of
> defining explicit control flow would be highly rewarding - since it would
> represent a necessary evolution beyond machine-understandable markup and
> entities. How entities are a constituent of higher level goals and
> processes is probably the real answer to better search. If not search, they
> would be a very interesting in terms of knowledge discovery - such as being
> to ask 'What happens at the Chile embassy [location]?' in Sam's example, to
> use just one permutation of many possible questions. Bringing all this to a
> scale such as the web would be very exciting.
> >>>>>>>>>
> >>>>>>>>> We at Tallyfy can help to define and implement Process markup,
> but we are one of many others. Is there a way that a project with some
> organisation can be spawned from this discussion?
> >>>>>>>>>
> >>>>>>>>> Thanks,
> >>>>>>>>> Amit
> >>>>>>>>>
> >>>>>>>>> On 9 Sep 2013, at 11:33, Martin Hepp <
> martin.hepp@ebusiness-unibw.org> wrote:
> >>>>>>>>>
> >>>>>>>>>> All:
> >>>>>>>>>> If you really want to embark into process modeling in
> schema.org, then you should first become clear about
> >>>>>>>>>>
> >>>>>>>>>> - whether you want to model processes in procedural fashion
> (explicit control flow) or a declarative fashion (modeling a set of actions
> and their pre- and post-conditions), and
> >>>>>>>>>> - whether the process models should be executable by a computer
> or merely documents for human consumption.
> >>>>>>>>>>
> >>>>>>>>>> Hundreds of researchers have worked on understanding how
> processes can be modeled in the context of information systems, and the
> least one can say is that
> >>>>>>>>>>
> >>>>>>>>>> 1. it is hard and
> >>>>>>>>>> 2. quick, simple approaches don't work or don't scale or both.
> >>>>>>>>>>
> >>>>>>>>>> See e.g.
> >>>>>>>>>>
> >>>>>>>>>>
> http://www2.informatik.hu-berlin.de/top/download/publications/fahlandlmrwwz_2009_emmsad.pdf
> >>>>>>>>>>
> >>>>>>>>>> for a brief overview.
> >>>>>>>>>>
> >>>>>>>>>> Without excluding others, I think it would make a lot of sense
> to involve
> >>>>>>>>>>
> >>>>>>>>>> Wil van der Aalst, http://wwwis.win.tue.nl/~wvdaalst/
> >>>>>>>>>> and
> >>>>>>>>>>
> >>>>>>>>>> Jan Mendling, http://www.wu.ac.at/infobiz/team/mendling
> >>>>>>>>>>
> >>>>>>>>>> in any such draft. They both spent years of their lives into
> understanding the challenges of process modeling...
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>> Martin
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>> On Sep 6, 2013, at 10:04 PM, Vicki Tardif Holland wrote:
> >>>>>>>>>>
> >>>>>>>>>>> I think a combination of Jason's suggestion of
> http://schema.org/ItemList and something similar to
> http://schema.org/Recipe would do the trick. The key difference is that
> you probably want to specify the step number instead of relying on page
> layout as parsers often discard the order of elements.
> >>>>>>>>>>>
> >>>>>>>>>>> Vicki
> >>>>>>>>>>>
> >>>>>>>>>>> Vicki Tardif Holland | Metadata Analyst | vtardif@google.com|
> 978-613-9630
> >>>>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>>> On Fri, Sep 6, 2013 at 7:17 AM, Tallyfy <hello@tallyfy.com>
> wrote:
> >>>>>>>>>>> "Process" sounds very promising as a purely top-level
> construct, because any serial process (not related to a "thing" but maybe
> with embedded references to things) can be wrapped and labelled as an
> actionable container. http://schema.org/Recipe is the same concept as
> this, but only relates to food recipes.
> >>>>>>>>>>>
> >>>>>>>>>>> We subscribe the Gates quote - "the future of search is verbs"
> and interpret it as machines able to understand not just content, but
> processes like "How to get a Chile tourist visa for British citizens" - an
> ordered list of steps. Rankings for processes are also different to content
> backlinks, which we are working on, as you could define pre-requisites (do
> this before doing this) and chain processes after (after doing this -
> continue with this).
> >>>>>>>>>>>
> >>>>>>>>>>> Could somebody help me propose this as a new item? I have no
> idea where to start.
> >>>>>>>>>>>
> >>>>>>>>>>> thanks
> >>>>>>>>>>> Amit
> >>>>>>>>>>> http://tallyfy.com
> >>>>>>>>>>> On Thursday, 5 September 2013 at 17:36, Sam Goto wrote:
> >>>>>>>>>>>
> >>>>>>>>>>>> Maybe an ItemList (or a specialized subclass, e.g.
> http://schema.org/Process) of http://schema.org/Action and its subclasses?
> >>>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>>>>> On Thu, Sep 5, 2013 at 9:33 AM, Tallyfy <hello@tallyfy.com>
> wrote:
> >>>>>>>>>>>>> The list may not be about a specific thing, but a process -
> which could include many things. For example - the list, "How to enjoy a
> great Saturday night in" might have a reference to a food - pizza AND a
> movie - as an entity, etc. Granted, the example isn't the best, but it's
> entirely unrelated to any specific thing.
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> In the composite scenario (which might not even have any
> linked entities) - I guess there might not even be a thing here at all,
> it's quite specifically a set of steps with an objective. For example "What
> to look out for when buying a house in London"
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> So to clarify, this isn't to enumerate objects or things
> into a determined order like "Top 10" - it's to define actionable things as
> steps - whether or not there's related entities.
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> A
> >>>>>>>>>>>>> On Thursday, 5 September 2013 at 17:24, Jason Douglas wrote:
> >>>>>>>>>>>>>
> >>>>>>>>>>>>>> Maybe a new subclass of ItemList?
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>> Aside: seems like ItemListElement should have a range of
> Thing so you could do structured lists (movies, steps, etc.).
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>> -jason
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>> On Thu, Sep 5, 2013 at 2:44 AM, Tallyfy <hello@tallyfy.com>
> wrote:
> >>>>>>>>>>>>>>> Hi everyone,
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> I run a startup called http://tallyfy.com
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> We've just been enrolled into StartupChile, and aim to
> launch within a few months using their help. Our homepage looks something
> like this:
> >>>>>>>>>>>>>>> https://dl.dropboxusercontent.com/u/14563542/tallyfy.png
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> What we do is allow anyone to embed knowledge as steps in
> a checklist or a process. Examples might be:
> >>>>>>>>>>>>>>>  How to bake a carrot cake
> >>>>>>>>>>>>>>>  How to change a bicycle tyre
> >>>>>>>>>>>>>>>  What to pack if you're visiting the Amazon rainforest
> >>>>>>>>>>>>>>>  My bucket list
> >>>>>>>>>>>>>>> The clearest and most obvious point to make here is that
> these checklists, when marked up via schema.org would be excellent ways
> to present answers to questions without people going through many pages on
> search engines.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> So I wanted to propose a schema for marking up a checklist
> (or a process).. If there is one already - could someone point me to it?
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> If we could understand that this is a "set of steps for
> doing something" - I think that would be very valuable, not just to search
> but for people looking for knowledge which is actionable, not just web
> pages. In other words, an actual set of steps marked up is more valuable
> than a block of content (usually using <ol> or <ul> HTML) which blends into
> a web page.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> We intend to do a lot more - you can measure how many
> people did a checklist, how long it took on average, reviews, etc. so
> perhaps those could incorporate into this schema.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> thanks
> >>>>>>>>>>>>>>> Amit
> >>>>>>>>>>
> >>>>>>>>>> --------------------------------------------------------
> >>>>>>>>>> martin hepp
> >>>>>>>>>> e-business & web science research group
> >>>>>>>>>> universitaet der bundeswehr muenchen
> >>>>>>>>>>
> >>>>>>>>>> e-mail: hepp@ebusiness-unibw.org
> >>>>>>>>>> phone: +49-(0)89-6004-4217
> >>>>>>>>>> fax: +49-(0)89-6004-4620
> >>>>>>>>>> www: http://www.unibw.de/ebusiness/ (group)
> >>>>>>>>>> http://www.heppnetz.de/ (personal)
> >>>>>>>>>> skype: mfhepp
> >>>>>>>>>> twitter: mfhepp
> >>>>>>>>>>
> >>>>>>>>>> Check out GoodRelations for E-Commerce on the Web of Linked
> Data!
> >>>>>>>>>>
> =================================================================
> >>>>>>>>>> * Project Main Page: http://purl.org/goodrelations/
> >>>>>>>>
> >>>>>>>> --------------------------------------------------------
> >>>>>>>> martin hepp
> >>>>>>>> e-business & web science research group
> >>>>>>>> universitaet der bundeswehr muenchen
> >>>>>>>>
> >>>>>>>> e-mail: hepp@ebusiness-unibw.org
> >>>>>>>> phone: +49-(0)89-6004-4217
> >>>>>>>> fax: +49-(0)89-6004-4620
> >>>>>>>> www: http://www.unibw.de/ebusiness/ (group)
> >>>>>>>> http://www.heppnetz.de/ (personal)
> >>>>>>>> skype: mfhepp
> >>>>>>>> twitter: mfhepp
> >>>>>>>>
> >>>>>>>> Check out GoodRelations for E-Commerce on the Web of Linked Data!
> >>>>>>>> =================================================================
> >>>>>>>> * Project Main Page: http://purl.org/goodrelations/
> >>>>>>>
> >>>>>>> --------------------------------------------------------
> >>>>>>> martin hepp
> >>>>>>> e-business & web science research group
> >>>>>>> universitaet der bundeswehr muenchen
> >>>>>>>
> >>>>>>> e-mail: hepp@ebusiness-unibw.org
> >>>>>>> phone: +49-(0)89-6004-4217
> >>>>>>> fax: +49-(0)89-6004-4620
> >>>>>>> www: http://www.unibw.de/ebusiness/ (group)
> >>>>>>> http://www.heppnetz.de/ (personal)
> >>>>>>> skype: mfhepp
> >>>>>>> twitter: mfhepp
> >>>>>>>
> >>>>>>> Check out GoodRelations for E-Commerce on the Web of Linked Data!
> >>>>>>> =================================================================
> >>>>>>> * Project Main Page: http://purl.org/goodrelations/
> >>>>>>
> >>>>>
> >>>>> --------------------------------------------------------
> >>>>> martin hepp
> >>>>> e-business & web science research group
> >>>>> universitaet der bundeswehr muenchen
> >>>>>
> >>>>> e-mail:  hepp@ebusiness-unibw.org
> >>>>> phone:   +49-(0)89-6004-4217
> >>>>> fax:     +49-(0)89-6004-4620
> >>>>> www:     http://www.unibw.de/ebusiness/ (group)
> >>>>>         http://www.heppnetz.de/ (personal)
> >>>>> skype:   mfhepp
> >>>>> twitter: mfhepp
> >>>>>
> >>>>> Check out GoodRelations for E-Commerce on the Web of Linked Data!
> >>>>> =================================================================
> >>>>> * Project Main Page: http://purl.org/goodrelations/
> >>>>>
> >>>>>
> >>>>>
> >>>>>
> >>>>
> >
> > --------------------------------------------------------
> > martin hepp
> > e-business & web science research group
> > universitaet der bundeswehr muenchen
> >
> > e-mail:  hepp@ebusiness-unibw.org
> > phone:   +49-(0)89-6004-4217
> > fax:     +49-(0)89-6004-4620
> > www:     http://www.unibw.de/ebusiness/ (group)
> >         http://www.heppnetz.de/ (personal)
> > skype:   mfhepp
> > twitter: mfhepp
> >
> > Check out GoodRelations for E-Commerce on the Web of Linked Data!
> > =================================================================
> > * Project Main Page: http://purl.org/goodrelations/
> >
> >
> >
>
>
>
Received on Tuesday, 10 September 2013 14:48:50 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:29:30 UTC