Hi all.
On Thu, Dec 13, 2018 at 2:02 AM Ziko van Dijk <zvandijk(a)gmail.com> wrote:
I just watched the showcase of December 2018, thank you for the interesting
contribution!
For those interested who haven't watched it, Ziko is referring to:
https://www.mediawiki.org/wiki/Wikimedia_Research/Showcase#December_2018
Thanks, Ziko! More below.
It would be great it further research could have a
look at
questions such as language choice.
Agreed. This has been a request by a few other community members as
well. One interesting question to address here is: can we characterize
language switching? More specifically: are there specific conditions
under which switching happens? This will allow us to answer questions
like: Are there specific topics that are covered in language x and not
y that trigger switching? Is switching a function of availability of
content or we can still see switching even when the content exists in
the 2+ languages the user is comfortable reading content in? ...
Diego started looking into this, and you can follow his future work at
https://meta.wikimedia.org/wiki/Research:Characterizing_Wikipedia_Reader_Be…
We will do more work in this space in coming 6 months.
With regard to have more insight in what readers want,
I struggled in the
past with two questions:
Regionally important content: Should a Wikipedia
language version
concentrate on regional topics, or try to cover a large variety of topics?
This is a good question, and as you stated, it is related to
understanding reader needs and some of the research in understanding
language switching behavior can help us understand this better.
Another aspect to keep an eye on is Denny's recent proposal for
abstract Wikipedia [1]. If that direction is picked up, we may have
more reason to emphasize on regionally important content creation
first.
Large or small articles: Some printed encyclopedias
had relatively few, but
large articles. Others segmented the content into many small articles.
(Think of Encyclopedia Britannica: Macropedia and Micropedia.) What do
Wikipedia readers want? Do they prefer to read about a larger topic in one
long, well structured article? Or several short ones, linking to each other?
This is an interesting one, too. There are at least two ways to
approach this question: study how Wikipedia readers learn (what it
means to learn needs to be defined) and then do a series of user
studies across languages and regions to find patterns and provide
recommendations for how to organize content with readers in mind. The
other approach, which I would love to see in action, is to break down
the article into many pieces and allow the reader to pick and choose
to create a learning experience for learning topic x. Then, learn from
the way readers learn. This will be building on Collection [2], Gather
[3] or other similar initiatives. Search data can also be valuable
here. (just to be clear: this is not something we're looking into
right now, but it's a fascinating area that if someone has bandwidth
and resources to look into, it can help us learn a lot.)
I could imagine that a reader who is interested in
information for work or
school prefers long articles that provide an in-depth approach in order to
became familiar with the overall topic (that is, what one would expect
traditionally).
We don't know if this assumption is correct: in fact, we have the
length of article as a feature in the study and it's not picked up as
a feature that defines this user group. What we know is that across
the 14 languages in the study, this group of readers have longer dwell
times on articles, they use the desktop platform, and they come to
Wikipedia in the afternoon [4].
The above being said, we can't say for sure based on the recent study
that this group of readers don't prefer longer articles because if the
longer article in the topic of their interest doesn't exist on
Wikipedia, they may have to work with the shorter article. It would be
great to have some user studies to understand this group and their
needs better.
And that "news" readers want to look up
something quickly,
in a short, simplyfing article.
This one we don't know. :) What we know is that across languages, this
was not observed as a consistent pattern (check table 2 in the most
recent paper [5]. for enwiki specific audience, check table 2a in the
first paper [6]: while 38% of the users motivated by media are coming
to look up a fact another 62% are there for overview or in-depth
reading.).
On Thu, Dec 13, 2018 at 5:58 AM Bob Kosovsky <bobkosovsky(a)nypl.org> wrote:
"Large or small articles." I've noticed this point of contention at the
outset of my Wikipedia editing. There are some editors (and presumably
readers) who want Wikipedia to look and function like a traditional
encyclopedia, with thorough articles reflecting well-written and thoughtful
essays that one used to find in encyclopedias. Those who know anything
about web design know that a long essay goes against the design ethos of
the web where some advise against webpages that require excessive scrolling.
We need to understand this better. What we see in the recent study is
that readers in countries with low Human Development Index read
Wikipedia more frequently in-depth (when compared to those in high HDI
countries). What we don't know is if the current forms that the
articles are written in is satisfying their learning needs or they
would prefer to read and learn using the same content but in different
representations. I shared some ideas in my response to Ziko how we can
learn more about this aspect of reader needs.
The bottom line is that I don't think one can or
should make a definitive
rule regarding these issues because different communities will want
different attributes and styles. To be sure, editors/readers should be
aware that such options exist and that Wikipedia style varies considerably
from article to article (and community to community).
Agreed, but I suggest we don't stop there:
* We should experiment with ways to bring editors and readers closer
to each other. I mention this in the discussion part of the showcase
as well: at the moment, the broadly available link from readers to
editors is pageviews on an article page, and perhaps some other
features. We can experiment with ways that can empower editors to
understand the audience of their articles better.
* We can think of ways to make the content and its representation less
rigid from the reader perspective. While each language community has
their own style of writing, we can experiment with allowing the reader
to pick and choose content and represent it in the way that is most
useful for their reading needs.
On Thu, Dec 13, 2018 at 6:31 AM Federico Leva (Nemo) <nemowiki(a)gmail.com> wrote:
Ziko van Dijk, 13/12/18 12:02:
Regionally important content: Should a Wikipedia
language version
concentrate on regional topics, or try to cover a large variety of topics?
This question is automatically solved if instead of focusing on
Wikipedia you do Wikisource. Wikisource will only contain texts
published in that language, such as local fiction and official acts of
local entities.
Correct, but the sources in a given language may or may not be about
regional topics. So even in the case of Wikisource, the question of
whether to focus on regional (geographically close) topics can be
valid. I /think/ in the case of Wikisource you can imagine that while
it's important to capture all of the possible sources of a language,
you may want to prioritize region-specific sources over others if you
have specific objectives.
Best,
Leila
[1]
http://simia.net/download/abstractwikipedia_whitepaper.pdf
[2]
https://www.mediawiki.org/wiki/Extension:Collection
[3]
https://www.mediawiki.org/wiki/Gather
[4] First bullet point on page 6.
[5]
https://arxiv.org/pdf/1812.00474.pdf
[6]
https://arxiv.org/pdf/1702.05379.pdf