[microformats-discuss] Re:ratings

[Tantek Ç elik]

On 7/27/05 8:27 AM, "Peter Jones" <soundrater at yahoo.com> wrote:

> Kevin Marks wrote:
> 
>> Peter, do look at hReview, which details a way of
>> expressing tagged ratings in a microformat:
> 
> Thanks for pointing out hReview - I have looked at it
> and there were a couple of reasons why I thought
> VoteLinks was more suitable/applicable:
> 
> 1. Reviews require quite a lot of effort to compile
> (when compared to ratings) so capturing ratings under
> a *Review* microformat could put some people off.

Why? Ideally systems shouldn't bother the user with the underlying
format(s).

In other words, the user is rating something or reviewing it, rather than
"writing an hReview".

It would be trivial to take the current hReview creator

 http://microformats.org/code/hreview/creator

and strip it down to a "rating creator" that only allowed you to rate a
resource, and yet, still output valid hReview markup, which could be
properly parsed/aggregated/indexed etc.


> So, stripping down the Multidimensional Restaurant
> hReview example provided in hReview would leave the
> restaurant url,
> 
> <a class="url"
> href="http://cafeborrone.com">cafeborrone.com</a>

Note that this simplification has thrown out some valuable information about
the restaurant being a business with a location etc.


> and the rating (which in this case is for food),
> 
> <a href="http://en.wikipedia.org/wiki/Food"
> rel="tag">Food: <span class="rating">18</span>/<span>
> class="best">30</span></a>
> 
> This may work fine, but is based on the tag assumption
> that an external tagspace is required to define the
> tag, hence the use of rel="tag".

Yes.  Using tagspaces for disambiguation is core to the rel-tag spec.

> All I am looking to
> do
> is indicate a criterion against which to make the
> rating of an external resource, and I doubt it needs
> to
> go as far as having a defined tagspace.

I applaud your effort to make it as simple as possible.

But when it has already been tried and failed, it's worth not repeating
others' mistakes (unless you have some reason to believe your attempt will
have a different result).

Arbitrary keyword/category/tagging systems where the keywords themselves are
not tied to any kind of definition or other context have been shown to be
quite noisy and not very useful.

Delicious showed that by linking a tag to a page that showed URLs with that
tag, it "defined" what was meant by that tag.

Flickr demonstrated the same with tags on photos.

rel-tag has generalized this pattern to arbitrary tagspaces (even user
tagspaces) and thus enabled folks to pick their tagspace rather than being
bound to the tagspace of the service that they are using.


> Could ratings be done more simply if structured along
> the lines of VoteLinks? So, for example:
> 
> <a rev="vote-for" href="http://ragingcow.blogspot.com"
> title="neat spoof">Raging Cow</a>
> 
> could be coded as:
> 
> <a rev="rating" href="http://ragingcow.blogspot.com"
> title="4outof5">Spoof</a>

As Ryan pointed out, it is undesirable to hide (or even partially hide) such
key user authored information as the 4 out of 5 rating.

This technique is essentially trying to squeeze more into a hyperlink than
can be done in a good way.

> I think this issue of using *rel* for tags (i.e.
> incoming tags) to say something about a post could be
> confusing if people want to use tags as criteria for
> rating an external resource along the lines of the
> *rev* attribute.

Also as Ryan said, xFolk addresses this:

 http://microformats.org/wiki/xfolk

> So it may be best if I avoid the idea
> of using a *tag* as the criteria against which to make
> a rating for an external resource, and just stick to
> establishing rating criteria of some sort without
> calling it a tag.

Yes, the simple case in hReview is to just offer an overall rating for the
item.


> Anyway, back to why I joined this list - are there any
> plans for a simple rating microformat that just
> records
> the resource being rated as the destination of a link,
> the rating, and the criteria against which the
> resource
> has been rated?

I believe hReview already addresses that general use case.

Do you have a specific use case or example you can point to?

Thanks,

Tantek