#microformats 2023-12-19

2023-12-19 UTC
gRegor, ehmry, [Murray], [KevinMarks], [tantek], marion, [aciccarello], [Caleb_Hearth], Loqi_ and Zegnat joined the channel
#
capjamesg Where can I find out more about the discussions going on around u-review-of and h-review?
#
capjamesg I have some thoughts to share.
#
capjamesg Found it.
#
capjamesg https://github.com/microformats/h-entry/issues/32
#
Loqi [preview] [barnabywalters] #32 Consider merging h-review properties into h-entry
#
GWG capjamesg: What h-entry type posts have a p-rating?
#
GWG Can you derive review from rating?
#
capjamesg Are you asking if the presence of a p-rating is an implied review?
#
GWG I like h-entry u-review-of h-cite/h-card/h-item
#
GWG capjamesg: I don't think of it that way necessarily, just a good question.
#
capjamesg I'm a bit confused.
#
capjamesg "There might be a case" doesn't feel like a good way to start an argument.
#
capjamesg I guess I'm unclear on the immediate why.
#
GWG Why about p-rating or u-review-of?
#
GWG My use case for eliminating h-review is reducing complexity.
#
capjamesg u-review-of.
#
GWG I never implemented h-review, but I quickly added experimental u-review-of support
#
capjamesg I implemented h-review :D
#
capjamesg Twice, with two consuming applications.
#
capjamesg I'm not against the proposal. With that said, I am curious about how many publishers would be affected by this.
#
GWG I'm not sure, but it is worth figuring out
#
capjamesg h-review wasn't mentioned in https://almanac.httparchive.org/en/2022/structured-data
#
capjamesg But h-entry was.
#
capjamesg I can see the argument that adding one property is easier than a whole new set of classes.
#
GWG capjamesg: And that is basically why I did it when the idea came up
#
capjamesg Hm.
#
capjamesg You've convinced me.
#
gRegor There might be more mf2 hReview out there than mf2 h-review
#
gRegor I would expect if we merge, we would still include a note that consumers should check for h-review
#
gRegor I have a draft blog post with some thoughts on that specifically for the movie reviewing session we had at IWC SD. short version is basically "check for h-review... if not found then check for h-entry with review-of"
#
gRegor Hope to have that published this week while it's all still fresh in my mind 🤞
#
GWG I have to read the notes on that
#
gRegor argh, I can't type today. I meant *mf1 hReview above
#
gRegor There weren't many notes, part of the reason I want to blog it
#
[tantek] yes gRegor that makes sense to me too
#
[tantek] I'd also like to figure out a way to transition or terminate h-review-aggregate
#
[tantek] Obsolete that is
#
[tantek] There aren't any known implementations though maybe there are publishers. So it's worth explicitly obsoleting to discourage future consuming implementations
#
[tantek] And then figure out what to do with mf1 hReviewAggregate which I believe is consumed by Google (they're the ones that asked for it in the first place)
#
gRegor I think capjamesg publishes the aggregates on Breakfast and Coffee
#
gRegor ah, interesting history. didn't realize that came from google
#
gRegor Yeah, h-review-aggregate on https://breakfastand.coffee/Fortitude_Coffee_-_Newington_Road_-_Edinburgh
#
[tantek] Google wanted to make it "easier" for them to show a search result snippet with the summary of reviews about an item. So rather than crawling actual reviews they wanted sites to markup their existing visible summaries. What could go wrong?
#
gRegor haha
#
capjamesg :)
#
capjamesg To remove h-review-aggregate, could we have a "u-review-of aggregation algorithm" which specifies how to calculate an aggregate of multiple h-entries with the u-review-of attribute?
#
capjamesg If aggregation is left up to the consumer rather than the publisher, we should be clear on how that works.
#
capjamesg I do publish h-review-aggregate, yes.
#
gRegor sum the ratings, take the mean
#
gRegor h-review's p-best and p-worst allow customizing the rating scale, but if they're not specified the default is 1-5, 5 being best
#
gRegor I expect those properties are pretty rarely published, 1-5 default usually works well
#
[tantek] Idk I think the simple algorithms are naive as to the real world abuses and biases of reviews
#
[tantek] For example as a consuming implementation I might want to only aggregate reviews from indieweb sites or perhaps also profiles on review sites that have green rel-me checkmarks linking to indieweb sites
[0x3b0b] joined the channel
#
gRegor Makes sense, but probably not something the microformats spec would direct on?
#
[tantek] correct, it's a reason why the microformats spec for h-review would NOT provide any guidance for how consuming code SHOULD (or SHOULD NOT) aggregate the reviews
trwnh joined the channel