Re: Review structure

On 16 June 2014 02:26, Dan Scott <dan@coffeecode.net> wrote:

> On Mon, Jun 16, 2014 at 12:13:34AM +0100, Alf Eaton wrote:
>
>> I've been adding microdata markup to some reviews recently, as well as
>> harvesting reviews from other sites, and have found that the current
>> reviewBody and ratingValue properties are often too coarse-grained to
>> properly describe the reviews.
>>
>> For example, this review of a restaurant has 3 separate aggregate rating
>> values ("food", "decor" and "service"), but they're just marked as
>> "ratingValue" with no way to distinguish between them using the microdata
>> alone:
>> http://www.zagat.com/r/kevin-rathbun-steak-atlanta
>>
>> Another example is this review of a film, which has three rating values
>> (not marked up as "reviewRating", in this case), each with its own section
>> heading and review text:
>> http://www.littlewhitelies.co.uk/theatrical-reviews/belle-26939
>>
>> A third example is the peer reviews of this academic article, which
>> contain
>> structure within the review bodies - "Basic reporting", "Experimental
>> design", and "Validity of the findings":
>> https://peerj.com/articles/435/reviews/
>>
>> I'd like to be able to extract all that information without having to do
>> extra HTML parsing and analysis after reading the microdata.
>>
>> One solution I can think of would be to add a "reviewSection" property,
>> having a "reviewSectionHeading",  "reviewBody" and "reviewRating", for
>> each
>> section of the review. There may well be a better solution, though…
>>
>> The AggregateRating class would probably need something similar (as in the
>> Zagat example above).
>>
>> Does anyone know if this has been discussed anywhere previously, or
>> whether
>> there are analogous "section" properties in classes other than Review?
>
>

> Hmm, interesting problem. If I was to constrain myself to existing
> schema.org properties, I would consider simply separating each review
> into three separate Review entities, using the existing "name" property
> instead of your proposed "reviewSectionHeading".
>

That seems reasonable. Would the individual reviews then be "reviewSection"
properties of one over-arching Review, with a shared author, date, review
subject, etc?

Alf