This is a QuickTopic space for discussion of the RVW metadata format for reviews. You can subscribe to the list (to receive new posts by email) using the button at the top of the page.

Alf,

Thanks for this, my first impression is that it looks great. I certainly intend to adopt it. One question at this stage: is the rating scale 0-100 or 1-100?

I think it is useful to have the zero option as it can be used to represent Unrated/Not Rated (as in Amazon.com). This option could be used by the creator of content to post what they consider to be the “correct” metadata (e.g. music genre) for it on their website and indicate that it is as yet unrated (and they do not consider it appropriate to rate it themselves as they created it although they would welcome others doing so). Also, reviewers could signal that they are, for example, reading a book and will review/rate it, but as yet it is unrated.

I briefly mentioned this zero issue, along with how I will be using the rating range in the following blog post: http://soundrater.blogspot.com/2003/11/ant...ng-range-0-100.html.

Anyway, thanks again.

Peter Jones.

Peter,

The zero option does sound like a good idea, but what would be the arguments for using that rather than leaving the rating out altogether?

One thing I have been keen on is for the creator of content to specify the metadata against which it is reviewed/rated (e.g. the genre of music). Reviewers/raters then provide their review/rating against this metadata; it gives the creator a strong incentive to get the metadata correct from the start.

This also minimises the potential for metadata variations caused by reviewers/raters creating such data. Consistency would help when aggregating reviews/ratings from many sources if the metadata specifying the item is the same.

For this to work creators of content would need to be able to release the metadata for review/rating. As I have been looking at ratings (which can be analytically compared) rather than reviews, I was thinking how this could be represented in the rating scale. One way would be for the author/artist/musician to use zero in the rating scale.

However, leaving the rating out of RVW reviews altogether would work fine; especially if “Not Reviewed/Rated”, “Review/Rating metadata created by author/artist/musician etc.” or even “Content available to be reviewed/rated” was placed in the review section. This metadata could be released by the creator on their website to get it into the public domain.

I note that you say “…the idea is that you can put whatever metadata you like to describe the subject of the review, including unique identifiers if possible, and aggregators will make use of it selectively as best as they can.” This may work, even if there are some variations in the metadata, and the whole idea of creators of content being responsible for metadata may be an unnecessary complication.

Also, there is the potential that raters/reviewers who do not know about zero representing “Unrated/Not Rated” may use zero as the bottom of the rating scale, thereby causing confusion, as their rating of zero would not mean “Unrated/Not Rated”. This would suggest it would be better not to use zero in this way, and the 1-100 option would be preferable.

I’m happy to try either option (1-100 or 0-100), and leave the “zero representing not rated/unrated” thought to one side if necessary.

Out of interest, do iTunes use 1-100 or 0-100?

iTunes has ratings from 1-5 (20-100) and 'None'. If you set the rating to none, the whole rating field disappears from the XML for that track.

With the possibility of creators of content providing the full metadata for an item, perhaps the review could include a 'metadata link' to a URL where the full data could be looked up later on. This would save including a lot of data in the feed itself.

Alternatively, if the item being reviewed has a unique identifier, the aggregator would probably be aware of various web services where the full metadata appropriate for each each identifier type (eg ISBN, IMDB, MBID) could be looked up.

Yes, a unique identifier for the subject of the review is the way to avoid the potential complications/confusions outlined below.

Also, as iTunes drop the rating field if you set the rating to none, it would appear best to simply leave the rating element out of RVW if it is “unrated/not rated”, as you initially suggested.

Bug report: the http://purl.org/NET/RVW/0.2/ URL is 404.

I had a root around purl.org and it seems what was originally a domain needs to be changed to a partial redirect. However, you're not allowed to delete a domain. Maybe I should just go for http://purl.org/rvw/0.2 instead?

I know I'm probably a lone voice out in the wilderness, but wiyth all the growing concern amongst parents over suitable content in books/movies/CD's for their children it seems unfortunate that there is no "Suitability" content rating in the Reviews standard.

I mean even movies have rating GP or PG etc etc. At least THAT should be worked into the standard as far as movies are concerned!

My two cents!

As far as I'm concerned, the main use of the metadata attached to a review is for identifying the subject of the review, particularly when a unique identifier is not known. In which case, a 'Suitability' rating wouldn't be appropriate, as it wouldn't help to identify a movie, for example. This metadata would be better attached to the subject in a central database, and requested when needed (when the review is displayed by an aggregator, for example), as it's not likely to change from reviewer to reviewer.