cdn.media.ccc.de/.../wikidatacon2019-1068-eng-Reinstating_Female_Artists_to_the_Cultural_Record_via_Wikidata_hd.mp4

Right! Hello, everyone!

The clock is running.

Thank you to Wikidata
for the opportunity to speak here today.

I will be talking about reinstating
female artists to the cultural record.

First off, what am I doing here?

I work for an organization called digiS,

which is the Forschungs-und
Kompetenzzentrum Digitalisierung Berlin

We are a state-funded body.

We're based in
the Zuse Institute in Berlin.

which is part of the high-performance
computing network in Germany.

We are involved with GLAMs,
with cultural institutions in Berlin,

helping them get
their analog treasures online,

digitized online,
making high-quality metadata.

And one of our major focuses
is also long-term digital preservation.

So, we assist them with all of that,

And as I said, it's state-funded,

and with a focus on Berlin,

so we work with Berlin institutions.

Our standard operating procedure

is that we take the metadata

that the cultural institutions
have produced

and that is ingested into
the Deutsche Digitale Bibliothek.

That's our main focus.

And then on into Europeana
if the metadata is appropriately licensed.

Not all the institutions
put out their metadata with CC0.

But that's our standard route.

But we're always looking
at new ways to present data,

because a lot of the museums
and cultural institutions we work with,

archives as well,

sometimes focus a bit too much
on putting the data on a website.

You've got to realize
that we're dealing with institutions,

sometimes very small institutions
in Berlin,

where there's just one
or two people working.

They're really happy
if they've just got a website.

If we start coming to them and saying,
"Hey! Start producing your metadata

on LIDO or Metsmart or whatever,"
they tend to have a bit of a crisis,

but that's why we've got
this funding program

to try and help them through it,

because we also would like
the data that they produce

to be usable for machines,

to really leverage the power
of the semantic web and link data.

Which is why we became interested
in looking at Wikidata,

because some of our partners
were actually using Wikidata already,

pretty much as a resource.

They weren't inputting any data,
but to make their whole...

collection management easier,
they would then collect to Wikidata

to save them the trouble
of inputting everything

and that's how we ended up
with the Berliner Malweiber,

as they are called.

So, Malweiber is a pejorative term.

It just means "Painting Women,"

and it was used to refer to women artists
who were not taken seriously

by the male-dominated academe
at the turn of the last century.

These were women who were painting,

but they weren't allowed
to study at any universities

or any higher education institutions.

They were allowed to maybe
go to painting schools,

but they were producing a lot of work.

Some of the artists
that we've got in the database

are early 1850's as well, but most of it
was around the turn of the century

So, this is Anna Bernhardi's
self-portrait

from 1891, I think.

In Berlin, the first time
women were admitted

to the Schule der Kunst was 1919.

So, it took a while
before they got any recognition

for the work that they were doing.

So, the Stadtmuseum in Berlin, in 2016,
had a temporary exhibition

called "Berlin -- Stadt der Frauen,"

where they tried to readdress
this imbalance a bit

and to showcase
a lot of the work that they have--

the Stadtmuseum has their artworks--

and to actually put on
an exhibition for them.

Of the 20 women who remained
the focus of the exhibition,

most of them are already quite famous,

but there were still quite a few
artworks in the museum

that are from lesser-known artists.

And that's why, as part of this program--

this is the website from digiS--

there was a program in 2016
or a project by the Stadtmuseum

to then digitize a further 215
of their paintings

and produce metadata to that.

So, that was the basis of our work.

So, standard operating procedure:

we announced the ingest
on the Wikidata Import website.

It was actually the old site.
They've changed the way now.

And happily...

...cleaned up the data
as much as they could in OpenRefine

because I was dealing with a dataset
that had been edited

by about three different employees
of the museum.

So, you notice immediately
how great it would be

if people would use controlled
vocabularies and stuff like that

because for materials and techniques
there were very often different words used

to describe exactly the same concept.

I did a whole lot of cleaning up,

checked for which artists
were already existing in Wikidata,

which had to be added new.

All of this was uploaded,
and then tragedy struck,

because for a lot of these artists
I had no external references.

I do not have any reference in the GND,
the German Authority Control.

And so, the only place I could enter
their data was in Wikidata.

They didn't exist anywhere else.

And Wikidata admin, quite rightly,
deleted 23 of the artists

that I had added, by saying,

"No source, no sitelink, no backlink,
no data to understand who they are."

I waited a while, then I canceled.

So, I engaged in a discussion
with this admin

and he or she very kindly restored
the items that had been deleted,

which gave me the opportunity to say,

"From my point of view,
I did a bit of calculating back then,

I may be wrong; this is a year ago now.

There were 15 million items.

Of those, a million had been deleted.

That's about 2% of Wikidata."

[Arguing], I realized today
that in the earlier session

we'd say that Wikidata
was having a resource problem. Good.

We need bigger machines.

But it would be interesting to say,

maybe the notability criteria
should be made even more inclusive

if it's clearly spam or vandalism--
yes, that's not an issue--

but there may be some items
that are a little bit on the borderline

and maybe Wikidata would be
an interesting place to store that data,

or Wikibase, if we're going to see
some other options

of what we can do with it.

So, a suggestion would be to set up
some sort of arbitration process,

because what annoyed me a little bit
about having my items deleted

was the fact that I wasn't really
given a chance to argue my case

in any way, shape, or form.

and it would have been nice if someone
gave me the opportunity just to say,

"Hang on! This is actually real data,"
and explain the situation.

On Wikipedia, we have pages very often
with a banner on the top:

"This page has some issues."

Maybe we could think
of something similar for Wikidata,

but I do not know
how realistic that suggestion is.

It was just something I thought
that might make people be more open

to putting their work into Wikidata
and not be too terrified

that their work is going to be
somehow deleted at some point.

I'll just remind you
what the Wikidata criteria are.

Let me scroll.

This is not my computer,
and I am left-handed,

so I'm struggling here a bit.

I might just...

...use my left hand!

Let me scroll. Yes!

Ah! Of course! Windows.

There we go!

So, if it refers to an instance
of a clearly identifiable concept

or material entity,
the entity must be notable in the sense

that it can be described using serious
and publicly available references.

So, this was part of our battle

in how can we get
a reference for an artist

who was excluded by design
from any formal record.

It's quite a complicated situation.

Now, how do I get out of that page?

Going back to the left hand.

Finding the mouse. Hello, mouse?
Anyone seen a mouse?

(chuckles)

Ah! There we go!

Let's go back.

So, at this stage, there are
still 19 artists without a GND,

so this is the German Authority Control.

So, these are all German artists,
or Berlin artists as well.

So, one would hope
that they would qualify for a GND ID,

but at this stage they do not have one.

I have artists, for example,
like Ida Maurer-Hahn...

(humming)

There we go!

...where I can really only just say
this much about her.

I know she's human, I know she's female,
and I know she's a painter.

That's all I know.
I'm not the domain expert.

This is data I got from the museum.

The museum is busy setting up
all sorts of other things,

so it's a little bit
of a battle for them too

to allocate resources to this,

and we're stuck in the middle of digiS,

trying to make sure that good quality
metadata is being created,

and doing our best
to ensure that it's done.

There we go!

And this is, for example, in some cases,

even though I may not have
the metadata or an external reference,

I have a painting,

and I can at least link to that painting.

And for some of the paintings
that are online at Stadtmuseum--

but I won't-- well, maybe
we can explain this here.

This is what the metadata looks like
for one of the paintings, instance of.

I have all the data that the museum
has about this artwork;

however, I have no online link
to any of this stuff.

I can't reference it

because it's in their collection
management system.

It's not available online yet,
so how am I going to do this?

There is hope. Do not despair!

The German Authority Control
is run by the German National Library,

the Deutsche Nationalbibliothek,

and they have a project running
called GND4C,

which is the GND for culture data.

And, in fact, I presented
on this topic a few days ago

at the annual meeting
of the expert group on documentation

of the Museums Federation of Germany,

and there were people
in the audience from the GND4C

who have now said they will help me get
at least these 19 artists who have no GND

to at least get their IDs,
so that we can start the process of--

as it has been said
in another presentation today,

the moment this data is online in Wikidata
and someone does a query

and says, "Hey, here's an artist,
but there's not much data about them.

Maybe there's a story to tell here,
and maybe we'd be interested

in doing some research
and to expand that."

That's been the amazing effect
of uploading a lot of this data

when there has been IDs
or there has been the ability to leverage

all the link data power,

that all of a sudden,
so much more information

has been collected about these artists
who in the past had really been ignored.

What are the lessons
we can learn from all of this?

How can we improve on past performance?

I think it's difficult to find
the right name to describe this.

There is such a thing
as the Chief Data Officer.

It's very much a commercial
and business enterprise sort of thing,

but what we noticed is that
in the cultural and heritage institutions

which have a personnel
and resource problem anyway,

we would like to have
a Chief Data Officer,

at least someone who is responsible there
for making sure that the metadata

is of a high standard
and that it's not just--

There's a role in Germany
called the data redakteur,

I don't know what the English
translation is. I could not find it.

But someone who is mainly responsible

to make sure that there are
no spelling mistakes in the data

or that it's of a relatively good quality.

But we actually need a bit more.

And a rose by any name
would smell as sweet,

so I don't really care what we call it.
I care more about what they do.

And what they need to do
is to make strategic decisions

regarding this data usage.

Someone actually in the institution
needs to sit down and say,

"It's important that
we get online because..."

I unfortunately missed
the presentation by The Met today

to see what they've done with their data
once they've uploaded it into Wikidata.

But if we can explain
to the institutions why it's important,

and they can see the reason for this,

then maybe they will be able
to make funds available

or resources available
to have someone to take this task

of actually realizing a museum
or a cultural heritage institution

is really a data provider.

They need the architecture
to provide data.

They need to know
who wants to use the data.

Is it digital humanist researches?
Is it just the general public?

They could have one central system

that can maybe feed into their website
and also have an API.

And one of the issues with the DDB
is they have an API,

so the metadata that we uploaded
into the DDB can be queried by an API.

However, you need an API key
before you do that.

That makes Wikidata attractive again

because I can just go to
any Wikidata SPARQL endpoint

and basically do my query.

And we need to ensure high-quality data.

That’s the beginning
of this whole process.

Just for the 250 artworks that I uploaded,
the data took so much work

to try and massage it into a useful form
that could be really great,

to ensure data quality
and that would be the basic things,

the uniform measurement system
and controlled vocabularies

as much as possible.

And all of this compliant
with relevant metadata standards.

We always bang the drum for standards,
because the moment you use standards,

you increase the ability
to connect your data

to other forms of data out there.

And then to avoid some of the issues,

because I can understand
that Wikidata also says,

"Look, we're not your data repository.

You can't just come here
and dump your data."

And the institutions, on the other hand,
say, "Look, this is our curated data.

This is stuff we spent years researching

and spent many employee hours working on.

We don't just want
to upload that to Wikidata

and have it overwritten
or deleted at some point."

So, I really like the idea of Wikibase.

The idea that an institution could run
its own Wikidata instance,

where we really need to have a look
and see if we can assist...

Where's the pointy bit?

...is in this bit here.

So, if you have an institution

that has its own collection
management system,

we need to find a way, write scripts,
and support it as much as possible,

to as automatically as possible get data
into their local Wikibase instance

or an instance run
by another government institution,

perhaps like digiS-- but I can't
make that prediction at this stage--

that could then be used by Wikidata
in terms of a synchronized

or a federated query.

And all of these things can, of course,
then use authority controls and so on.

But if the institutions can be encouraged
to set up their own Wikibases.

10-15 years ago, very few museums
had their own websites.

I would hope that in five years

more institutions
would have their own Wikibase

or be prepared to use a Wikibase.

And, of course, it gets interesting

when the moment you input it
into a system like that,

issues with the data will arise

that you can hopefully feed back
into the collection management system

to improve their data in general.

So, in other words, something
like a DIY Authority Control

for the institution
or even just a controlled vocabulary.

If they just had a resource
within their institution

where they could say,
"We're going to choose this term,

we're not going to go through
a long standardization process.

Even if five years down the track

they decide it's the wrong term
and they want to change the term,

at least I have a clear idea
what I'm talking about."

It’s a start.

But I have to say
that this suggestion that I made

was also met with quite
a lot of skepticism at this conference,

because I think the people in the audience
have also had bad experiences

trying to set up these systems.

And there's the saying:

"To err is human, but to really screw up,
you need a computer."

So I can imagine if you've got people

trying to set up their own Wikibases,
their own control vocabularies,

a lot of things are going to go wrong,
but it's a learning experience.

And we hope that something good
will come out of it in the end.

This is something that was referenced

in the Wikibase inspirational
presentation this morning.

The GND is now cooperating with Wikimedia

to actually see how Wikibase
can be used for it,

because it's part of the whole GND4C,
opening the very strictly controlled

for libraries resource
of the Authority Control to now say,

"Hey, other GLAMS, when you use galleries,
libraries, archives, and museums,

also want to use this resource"

GND4C is trying to see
how they can actually make that happen.

And one of the things
they're exploring is Wikibase.

And just the last point to say,

if you're in Berlin
on the 5th of December,

and you'd like to come
to the digiS Annual Conference,

where every year our partners present
their projects, it's fascinating.

We've got a very broad spectrum.
You're welcome to come along.

Details there in German on the URL.

Thank you for your attention.

(applause)

And I'll just point out, if you want,
there are some interesting readings.

This never loads for some reason,

but it’s actually an article
by Linda Nochlan

about why there are no great
women artists from 1971.

It was republished
a few years ago, online.

The link there will work.
It's a fascinating article.

Invisible women: Exposing Data Bias
in a World Designed for Men,

also fascinating.

And this is a book about Die Malweiber.

Right!

(person 1) Can you show that...

- Mm-hmm. Yup.
- (person 1) ...[in writing]?

Questions?

(person 2) So, how did you initially
find those women?

Because they are in a museum
database, I guess.

How did they end up
in the museum database?

[inaudible]

So, the museum has artworks
by these artists,

so they basically got them
in a storage facility somewhere,

and they have it
in their collection system

saying this is this painting
by this and this artist.

So they've got that data.

(person 2) Then you solve
a notability problem, right?

It would if it was online,
but it's not online.

This collection resource--

(person 3) [inaudible]
been made public [inaudible].

Yeah, I can't--

(person 2) That's not needed,

like if you referenced
a book, that's also good enough.

(person 3) You could reference
an offline catalog.

Yeah, but again it's a question of, well,
I don't even have an offline catalog.

I could maybe say it's the museum
or whichever collection...

(person 2) I was going to say,
I understand it's a small museum,

and maybe they don't have
a very sophisticated--

Well, Stadtmuseum
is not small, but, yeah. (chuckles)

(person 2) Maybe
they don't have the resources

to really put up something
sophisticated metadata-wise,

but you could literally have them
publish even a press release.

Whatever mechanism they have
of pushing data online that is by them,

that is by the state museum,

that can say, "Here is a list of women
who were systematically excluded,

and yet, here is what we know about them.

And we say this as the state museum."

That is a source.

This was exactly the problem.

So, they did have
this exhibition that's online,

but that's only 20 of, I think,
50 women artists.

(person 2) But they can say that
about all of them.

That's why I think Wikidata
also relaxed the rules a bit,

and let me keep that data in,

because I had it all up
in the Wikidata import page.

But as I'm trying to say
with Ida Maurer-Hahn,

I don't have any external reference,

and I hoped, though,
just getting the GND IDs,

we'll have an opportunity to...

(person 2) I just wanted
to mention, this is Andrew,

and his talk was about The Met,

the one I mentioned
to you about this morning.

- So you two should talk.
- Fantastic!

(person 2) Because he's also doing
interesting things with museums,

So, I just wanted
to make sure you connect.

And we should publish
some of the artworks on Wikidata

to make the data more complete.

Well, I should also point out
that some of the things we uploaded--

So, the Stadtmuseum
is very focused at the moment

on the Humboldt Forum,
which is being set up in Berlin.

Of these artworks that we digitized,

some of them have now
run into rights issues,

some of them were taken offline again.

The moment we're digitizing this stuff,

half our work is really dealing
with legal issues

and trying to explain to the institutions.

Because even if an artwork
is in the public domain,

when it's being digitized or photographed,

certain rights are then
ascribed to the person.

(person 2) But to be on Wikidata,

it doesn't have to be
in the public domain.

No, but it has to have
a fairly unrestricted license.

Not on Wikidata, sorry,
but if I'm talking about the actual image.

- Sorry, I thought you meant the image.
- (person 2) I was talking about Wikidata.

The metadata itself, that's all
been put online. CC0, no problem.

(person 4) Hi! Your sources
don't have to be online.

So, if they have a printed catalog,

whether it's an exhibition catalog,
or a catalog of a whole collection,

you can create an item
about that publication

and then use a stated in property

to cite that.

That's the problem.
Not all the artists are in the catalog.

They do have a catalog,

but it's only 20 of 50,
I think, women artists.

That's the whole issue.

They selected out of their whole
collection of artists,

they only selected 20 of those artists.

So, there's still no written
record of them anyway.

Like I said, I mean, hopefully,
if within the next few weeks

I can get the GND IDs
of the 19 missing artists, then...

(person 4) And further than that,

if you can state
that a painting is in a collection,

and an artist is the creator
of that painting,

or indeed the subject
of the self-portrait painting,

then I think that satisfies
the requirements

for inclusion in Wikidata.

It seems to so far.
No one has deleted them yet.

Because that was that problem too.

The reason why those things got deleted

is there was a gap of a few months
between when the artists were uploaded

and when I could complete
uploading the paintings.

But, yeah, since all the paintings
have been linked now too.

But I don't have any of those
like Ida Maurer-Hahn.

They're just statements at the moment
but without any references.

Hopefully, that will change soon.

(person 5) Hi. Thank you for your talk.

I have a question concerning
the GLAM institution you mentioned.

What do you think
are their learnings on this project,

because [inaudible]
also have a lot of GLAM work,

and then for me it's interesting to see
what the institution is thinking

about a collaboration
with Wikidata for future work.

I mean, you sound a bit disappointed.

This is exactly the part where I need...

I need to be able to show them
what they can do with the data online,

because at the moment
it's a battle for us

even just to get them to deliver
structured metadata.

So we work a lot with LIDO.

That will be passed on
to the Deutsche Digitale Bibliothek.

It's a huge battle to get any data
into LIDO, for a start,

and then we upload it to the DDB,
which has its own issues.

Then with Wikidata,
I've sort of tried to show--

We actually had Jason Evans two years ago
give a presentation to say,

"This is what you can do.
This is what we've done in Wales."

But every city has got
its own speed and tempo.

Hopefully, out of this Stadtmuseum
with their data,

we can start doing
some interesting things,

find some applications
and other institutions

that can maybe upload their data.

But at the moment,
the museums, the GLAMs--

it's just not on their radar,

because they're so occupied
with fundamental existential issues

that, you know, linked data
is for them like,

"What the hell are you
wasting my time with that for?

I'm going to make sure
I can pay my employees,

of which there's only one, for next year."

But we try.

That's our role as we see it too.

This whole funding program
is supposed to kind of...

I mean, the other major thing
is long-term digital preservation.

We said, "Look,
you can digitize this stuff,

but having it
on a hard drive is not good.

You've got to have it
into a proper system,

where, in 100 years, someone
can actually access this resource."

(person 5) I feel your pain
about the relevance

because I ran into a lot
of these same issues of the Wikipedia,

especially the German language Wikipedia
has these really strong relevance criteria

that are absolutely observed

and tend to exclude
a lot of important information.

Would it help to change
the relevance criteria?

Would that at all be possible?

Because I'm afraid, of course,
that people will be afraid

that then "unnotable"--
whatever that means-- things

would manage to get in.

Well, this is it, I mean...

(sighs)

I think the moment someone makes
an effort to upload something,

it becomes notable to a degree,

and I think the criteria at this stage,
though, are actually quite flexible,

but I do think you run into that issue
of there are notability criteria...

There is a process
that has been put in place,

but it still depends on
what an admin does.

At that time, when I got in touch
with that admin,

he or she had deleted
150,000 different items.

So, there is a certain degree
of arbitrariness in the whole process,

because at least I could
convince him to put it back on,

but I know from the Wikidata mailing list

there's quite a few people
who ran into the same issue

where stuff has been deleted.

So, I think the solution is to go
the route, as I understand,

that Wikidata is proposing to
of Wikibase installations.

They're making it a lot easier.

The Wikibase Docker
is a lot easier to install.

I tried to install it about
a year ago and struggled,

and I managed to get
an instance up and running

and actually usable very easily
with the Docker container.

And I think that's the way to go.

If you can get the institutions to see
how easy it is to set it up.

15 years ago, they didn't have websites.
Now they have websites.

And I would hope for Berlin, at least,

that we can maybe get
a Wikibase instance going

and get a few museums
to upload their curated data,

and then see how
we can feed that into Wikidata.

And then I think everyone's happy,
I would hope, but yeah...

(person 6) I'm going to ask a question
you're unlikely to be able to answer.

But I'm going to ask it anyway.

I'm curious about the women
whose artworks haven't been collected

by the institutions

because it sounds to me

like there's a large group
of women artists,

some of whom won't have
had their paintings collected

and don't appear to be
in the record anywhere.

As a result, is there any movement
or any research being undertaken,

or anything you're aware of
to try and find these women

and get them into the record?

It’s like an extra step.

No, I think, this is the thing.

We can only start with the museums
what they've got, basically.

The Stadtmuseum gets a lot of donations,
as well from people around,

and then if maybe
something turns up, they go,

"Well, we don't know who this artist is,
and we can try to [track it back]."

That's the only serendipitous way
it's going to happen.

But I think Berlin and Germany
with its history

has lost so much artwork through bombing
and destruction, whatever, too,

that I think there's a lot
that has been lost permanently.

(person 5) Okay. Thank you.

(person 6) There's a group--

I can't remember the name
for the life of me.

but they actually do a lot of work
with Italian artists.

There's also work done in the UK

to kind of bring back women artists
back to the front.

And so, I think all these
local initiatives

might need to start working together

to do something global,
to push more for it.

Maybe that would do something more.

(person 7) I wanted to inquire
if there's a right direction to go

to animate institutions

to take part in making their data linked,

or open, in some kind of way available.

Should there be
other institutions approaching--

Jealousy, it has to be jealousy.

That's the only way you're going to do it.

(person 7) I see. (chuckles)

You need to get one institution
leading the way and people saying,

"Wow! Look what they can do
with their data! We want that too!"

I think that's the way web servers
got going really,

and I think that's the only way
linked data is going to work

is to make them jealous.

(person 7) So, you think
that the institutions

that are not having this great
linked data Wikibase instance stuff

are going to say, "Yeah, we need that!"

and then are trying to find some people
that can set it up for them.

Or should they approach other institutions
that can help them with it?

Or which is the right direction
for them to go?

For a starter, you've got to realize
that data is a process.

It's never finished.

It's always something you’re working on,

and we would encourage any institution
to take the first step,

but only to get cooperation.

So the Stadtmuseum Berlin
has helped a lot of smaller museums

because then they say,
"Look, we've got this stuff,

we need to do something with it."

And then they'll share it with
the Stadtmuseum, so to speak,

and help it get online.

But we really have institutions

where there's just two working on it
in there, part-time.

It's going to take a while
before they get on that trajectory,

but I would hope that
as the processes become better,

as Wikibase gets easier to use,

and as people get to see
what the advantages are of it.

But I really think it's a process

that you're always going
to have to support,

and it's something you're always going
to have to make clear

to the funding institutions,

that it's not something they can just
throw some money at, and then it's done.

It's going to need permanent funding.
Libraries need permanent funding.

Anything to do with digital resources
also need permanent funding, I think.

(moderator) Sorry, I think we have
just run out of time for questions.

I encourage you to keep talking
after the session,

and thank you so much to our speaker
and all the attendees.

Thank you.

(applause)