(cc-ing the designers, since I'm not sure who all is on mobile-l and who
isn't...)
Yesterday, some folks from design + me did some brainstorming around what
our first set of mobile contributions to Wikidata might look like. As a
reminder, at Q1 planning, we decided on 2 high-level flavors of features in
this area, both of which, thanks to Moiz, have horrible names :)
"Wiki-Tinder" was contributions that don't involve any freeform text input
and are just tapping/swiping – for example, tapping to tag a person's
gender. "Wiki-Twitter" was features that do involve some short freeform
text entry – for example, filling in the short descriptor field to describe
an article in 5 words or less. In this session, we focused on the
Wiki-Tinder side, since the ideas in that bucket were still pretty murky.
We decided on a few parameters up-front:
1) As with the mobile web upload workflow, we want to efface the difference
between all our sister projects so users aren't confused. So all of the
ideas below involve being on a Wikipedia article and being prompted to add
information that helps *Wikipedia*, but actually results in a Wikidata edit.
2) We want these contributions to be meaningful, both for our projects and
for the user – luckily, adding more structured data to Wikidata, no matter
how seemingly simple, will provide a tremendous benefit to all our projects
down the line. We just need to make that clear for our users :) So, we
decided that it's okay to test out some contributions that don't
immediately show up in the body of the article, as long as we can convey
their purpose to the user somehow (e.g., "Thanks! This will make searching
Wikipedia better in the future"). This is analogous to Yelp and Foursquare
prompts ("Is this restaurant good for vegetarians?") that don't immediately
get reflected in content until a certain number of people have answered the
prompt.
3) Since our yearly and quarterly targets are set to raise *new mobile
active editors *(users who register on mobile and go on to make 5+
edits/month on desktop or mobile), we want these games to generate edits.
So while it will be cool in the future to explore features that aggregate a
bunch of different people's inputs and only present information after X
number of responses, we're going to hold off on that for now and focus on a
one-to-one user-contribution model.
Here are the ideas we came up with:
1) *Adding Wikidata properties.* Over half of all items in Wikidata lack
basic information about what they are – all we have is the title of the
item.[1] When a user is viewing an article that has an entry in Wikidata
but no other information, we can present a CTA asking them to choose from
the common top-level Wikidata properties: person, place, event,
organization, etc.[2] We might explore doing this just once, or asking
people to continue categorizing an article several times.
2) *Refining Wikidata properties.* This is similar to the Wikidata games
that Magnus Manske came up with.[3] Once we know the high-level property of
a Wikidata item (whether it's a person, event, organization, etc.), we can
keep adding sub-properties to make that item more detailed. To do this, we
can present a CTA for users to add additional properties – e.g., if they're
on an article about a person, ask users "Does this person have an
occupation?" We'd want to build something generic here, in case we start
running out of items with property X and missing sub-property Y, as Magnus
began to ;)
3) *Wikidata quiz.* This is more about the presentation of 1 & 2 – one way
to think about how we show the CTA is as a quick quiz item for people
who've scrolled down past a certain point on the page.
The designers will start making some wireframes to make all this stuff a
little more concrete – we can go over them as a team at our next planning
meeting (next Wednesday the 16th) and hammer out more of the details. My
hope is that we can get some kind of prototype of 1 or 2 of these ideas –
even just a static set of pages that people can tap through – in order to
do some quick in-person testing before knocking together an alpha version
of the most promising candidate(s) in the sprint after next (last week of
July). If we play our cards right, that means we might have something to
showcase (and test with!) at Wikimania :)
1.
http://ultimategerardm.blogspot.com/2014/07/wikidata-items-with-no-statemen…
2.
https://www.wikidata.org/wiki/Wikidata:List_of_properties
3.
http://tools.wmflabs.org/wikidata-game/
--
Maryana Pinchuk
Product Manager, Wikimedia Foundation
wikimediafoundation.org