Re: Comment versus UserComments

Hi Daniel,

On Tue, Jan 17, 2012 at 7:14 PM, Daniel Dulitz <daniel@google.com> wrote:

> I wanted to raise an issue about how to represent comments (e.g. on blog
> posts).
>
> There are many subtypes of CreativeWork, but Comment is not one of them.
> Perhaps it should be?
>
> Instead, it seems like comments are to be represented by UserComments,
> which is a subtype of UserInteractions. But apart from UserComments those
> types appear to be for aggregates, not for an individual
> comment/tweet/like/etc. The type names are plural and don't really fit for
> non-aggregates.
>

I agree that Comment would be a relevant type to be added to schema.org.
There was a similar discussion on the previous mailing list where this
topic was discussed, in particular the confusion around markup of the
aggregates. I'm pasting some of the conversation below. See also the
whole http://schema.org/UserInteraction
inconsistency problems
thread<https://groups.google.com/forum/#!msg/schemaorg-discussion/N7u4Z8356Ao/JIiFO0WWNF4J>
.

Note that since its launch in January 2011, Drupal 7 exposes each
individual comment and the aggregate number of comments in RDFa using the
SIOC vocabulary, so I'd love to see a schema.org equivalent for both
individual comments and the aggregate number of comments for a given (blog)
post.

Steph.



---------- Forwarded message ----------
From: Stéphane Corlosquet <scorlosquet@gmail.com>
Date: Thu, Jul 21, 2011 at 2:03 PM
Subject: Re: http://schema.org/UserInteraction inconsistency problems
To: schemaorg-discussion@googlegroups.com


Hi,

On Sun, Jul 17, 2011 at 1:30 AM, Lawrence Woodman <lwoodman@vlifesystems.com
> wrote:

> **
> I have been looking further at http://schema.org/UserInteraction and find
> the layout
> to be very odd and was wondering what others thought?
>
> The first problem is illustrated well with the 'interactionCount' property
> of
> http://schema.org/Organization.  This property is specified as Text and
> the note gives
> examples of text such as '5 UserComments', But points to http://schema.org
> /UserInteraction
> where the examples use the format: 'UserComments: 78'.  So the plain text
> examples are using
> different formatting.
>
> The second problem is that http://schema.org/UserInteraction is a
> vocabulary which has
> various specific types such as: http://schema.org/UserComments.  But it
> does not seem at all
> obvious which property is meant to hold the value for this.  The examples
> are no help as
> they completely ignore that UserInteraction, et al are vocabularies and
> just specify plain
> text examples for the 'interactionCount' property of an Article.
>
> I think UserInteraction, would be better if it had properties for
> UserBlocks, UserComments, etc
> This would be far more consistent with the rest of the vocabularies and
> would look similar in many
> ways to http://schema.org/AggregateRating
>
> Has anyone else noticed this problem or got any further insight into its
> current use and how it could
> be improved?
>

I agree with you Lawrence, I also find the concept of UserInteraction and
the way the number of comments are defined unintuitive.

What stroke me first is the example available at
http://schema.org/UserInteraction:

[[[
This article has been tweeted 1203 times and contains 78 user comments.
  <meta itemprop="interactionCount" content="UserTweets:1203"/>
  <meta itemprop="interactionCount" content="UserComments:78"/>
]]]

A couple of problems here:

1. the types UserTweets and UserComments are embedded in the content
attribute with their value, so if I wanted to extract the number of
comments of the article for example, I would have do extra custom
processing to explode "UserComments:78", in order to get the type, and the
value for that type (with no property between the two?). The extra
processing required is not in line with the microdata or the RDFa specs at
all. For proof, you can try this snippet at
http://foolip.org/microdatajs/live/ and note that the value 1203 and 78 do
not get extracted in any way.

2. Microdata and RDFa both encourage reusing the existing content of the
page, so this example would be more in line with the specs by wrapping each
number with a span element:
[[[
<div itemscope itemtype="http://schema.org/Article">
  <span itemprop="name">How to Tie a Reef Knot</span>
  by <span itemprop="author">John Doe</span>
  This article has been tweeted <span itemprop="userTweets">1203</span>
times and contains <span itemprop="userComments">78</span> user comments.
</div>
]]]
which allows extraction of the number of tweets and comments following the
specs.


The definition of http://schema.org/UserComments is also strange: "User
interaction: A comment about an item." so according to this
definition, UserComments is not the number of comments as the example in
UserInteraction lets us think, but rather an actual comment left by a user?
That's not inline with example of UserInteraction.

Back to http://schema.org/UserInteraction, it says "A user interacting with
a page", and inherits properties such as:
- attendees (the people who attended another user leaving comment?)
- performers (the people who were involved in writing the comment?)
- duration (the duration a user took to leave his comment?)
- location (the location the user was at when he left his comment?)
I'm being slightly ironic here, but my point is to show that
this modeling is overly complex and not something people will implement...
all what is needed here in a property to specify how many comments (or
tweets, or blocks, downloads, likes, etc) a given page/article has.

My suggestion to solve the above issues would be to drop the
type UserInteraction and turn the existing types UserTweets, UserComments,
etc in properties belonging to a generic type like CreativeWork.

Steph.