Melanie Conroy – Networks In Literary History: The Salons Project

Melanie Conroy – Networks In Literary History: The Salons Project

[Melanie Conroy] So um, I’m here with the
Salons Project. We’re apart of Mapping the Republic of Letters at Stanford University.
And so for people who maybe don’t know what the salons are, they’re a big part of enlightenment
studies. So basically private meetings, usually at very elite women’s houses. Not always aristocrats,
but usually very you know the wives of wealthy bankers, or something like that. So it’s part
of a set of private institutions through Europe that were central for history of science for
literary history, theatrical history, but people tend to work on them in their own field.
So you know I’m in literature and you know we have our own work on the Salons, but we
don’t interact much with people in the history of science for example. So that’s one of the
big things that Mapping the Republic of Letters is working for, working towards. So Mapping
the Republic of Letters, the Republic of Letters is actually quite a vast thing. This is just
a screenshot of the visualization of letters moving throughout Europe. And the Republic
of Letters extended – it’s a scholarly network. Scholars use to send letters to each other
as one of the main means of communication, before the admin of things like journals,
obviously things like email. And so we can track the movement, for example scientific
information through letters. As one scholar would forward information that they received
from another scholar about any sort of new discovery. My part of Mapping Republic Letters,
is called Procope. After the first cafe in Paris. So this a not, not doing any visualization
of of Voltaire and Rousseau in the cafe of Procope, but it’s this legendary cafe. It’s
still exists today by the way. Where people got together and had philosophical discussions
that lead to much of enlightenment philosophy. So, Procope is a collaboration of Dan Edelstein,
myself Melanie Conroy, Maria Comsa, Claude William, and Chloe Edmondson. And we’re working
on the demographics of correspondence of major enlightenment figures. So it starts from one
kind of big, big man version of history. Where you’ve got the big man of, you know people
like Voltaire, who are very renowned philosophers. Thought what we want to do is dig deep into
their social networks, so looking at the people that they wrote to. Obviously in those networks
we find other people Rousseau. But we help- We find historical actors about whom we know
very little. We’ll find Mr. Brown of something. And you know we can guess he’s probably English,
he’s probably a guy. But that’s about all we know. What we did with Procope was to start
with the 18th century French correspondents to make the number a little more manageable.
Because we started with the entire data set of the electronic enlightenment project. Which
is over 7,000 correspondents. And it has information about the people occupation, and birth and
death date. Where they were born, where they died. So it’s a really fundamental resource
for enlightenment studies. But we focus just on the French people within that very large
data set. And what we did was to place individuals in what we call Things, Knowledge, Networks,
and Social networks. They’re not really networks because we don’t really know much about the
internal structure of the networks, they’re kind of more like a bucket of people or something.
Just sort of a grouping we’re sure, pretty sure that these people knew of each other.
For example, we have all the chemist together probably the 20 people working on chemistry
in the 18th century heard of each other. But we don’t have documentary evidence, such as
we would with letters for example we have people writing to one another. And so we started
from these very large networks like sciences, and then we broke that down, you know you
can have sciences and then just mathematical sciences in order to be able to group together
all of the people with a scientific interests or just the mathematician, just the chemists,
and then for letters again there’s philosophical letters. Letters is really anyone who writes
books. And so you could have religious letters, antiquarian, these sorts of things. And then
for social networks we really have to go with the biggest groups possible. So we arrive
at these categories Elite, Aristocracy, and Military. Elite and Aristocracy overlap to
a large extent, especially in France. But Elite is basically famous people, renowned
people, people who would be accepted at the most important salons. Aristocracy is just
a matter of births. So there’s plenty of Aristocrats who pretty obscure and you know couldn’t just
go to the cafe Procope and have a- expect a really warm welcome. And then things like
military, because we’re pretty sure that people within the military were more likely to know
one another. We didn’t include things like just being Catholic in France. We felt that
if you’re a protestant in France you might be more likely to know other protestants,
but just because you’re Catholic in France, you know 99% of the population is Catholic
in France, you probably don’t know many of the other Catholics. So the project leads
on just as the salons part. So I’m Melanie Conroy and I work on the salons and the academies.
Things like French academy and the Royal Society. But on salons I work with Chloe Edmonton at
Stanford. And our project is on the demographics of working in salons. But we did a pilot project
as a part of Procope on Parisian salons. So really just the 18th century the Enlightenment
and the very early 19th century to get a picture of you know, what happens with the French
Revolution. How does french high society change before and after the French Revolution? You
know what does it look like during the Enlightenment? Versus the early 19th century in the Neapolitan
Period. So we stated out with this idea that we were going to do you know straight up network
analysis. The big problem with that is that we have a number of sampling errors. Obviously
most of the time we are relying on biographies of Stella yang, the woman who constantly ran
the salons. So we’re constantly gonna find that they’re incredibly important in social
networks because we’re drawing from her biographies. We also found that the connectors between
networks. So there you’ve got the Princess Matilde and also Val Sack. But we mostly had
famous people, and very high rank people. Very high rank nobility. Famous writers, that
sort of thing as connectors between salons. Probably the reason fot that is that if you
have a salon you’re gonna claim, for example the Duke of Wellington comes to your salon,
if some obscure guy comes to your salon you’re probably not gonna write it down. And so,
we’re pretty sure we got a very bad sampling probably there. So what we decided to do is,
yes, use those looser concepts of networks. So, as an example, the salon of Madame Geoffrin
is one of the main meeting places for the Filosa. And, there’s a lot of controversy
around her salon, because salons in general were very aristocratic. But she had a number
of connections to the banking community and there were also some number of philosophers
at her salon. And it’s really important in those types of days to get a sense of whether
the philosophers, people like Rousseau or Voltaire were really like interacting more
with the aristocrats going to salons, mainly to get money, to get patronage. Or for entertainment
purposes. Or if they were actually having philosophical discussions that they would
publish in their works. So this is a fictional reconstruction of the salon of Madame Geoffrin.
You see all of the important Enlightenment figures there. It’s very heavy male. You see
Geoffrin over with the bonnet and the blue. And in the back you see a bust of Voltaire.
This painting was done by Lemonnier and its actually – he was someone who attended the
salon. But this is completely not historically accurate for some reason. It’s a complete
re-imagine and a fiction. Voltaire was dead at the time, so we fortunately put him as
a bust, but there’s so many other reasons that people can’t always be in the same room.
There’s people who never met each other. People who maybe didn’t go to her salon. and yea,
this is really one of the most iconic representations of the salons, with all the philosophers sitting
around, ready to discuss, or in the case listen to the reading of the play. So this is an
incredibly cleaned up version of our data. You can see obviously we pulled out the persons
full name which is not a trivial problem for French aristocrats since they often have many
many names. Looking at things like gender, nationality, which we simplify incredibly
to Italian and Germany- and I mean German- just to make it easier to do the analysis.
And then again, the knowledge networks. Letters and literaries, so that we search for everyone
in letters, or just the literary people. More easily things like political economy, which
is a lot of different aspects of what we now call the social sciences. And then, social
networks, elite government, aristocracy. And you know government is very precocious as
well because diplomats in there, any sort of high ranking officials because we, you
know we had very long discussions about this, but we don’t really feel like diplomats only
spoke among themselves. We feel like they probably knew other people who worked for
the government. We have another network for just servants. Because really felt that within
the professional networks you’re not gonna have the butlers hanging out with the butlers
and then the people who work in the kitchen hanging out in the kitchen stuff. It’s sort
of all the domestics would be meeting and talking to one another. So the attendees of
salons we code them for gender, interests (the knowledge networks), academic affiliations.
You know are they in one of the academies? Social networks and professional networks.
And here is some really quick data on the salons of Geoffrin. This is an example. So
you can see that you know there’s a fair representation of women. Some- Some historians argue there
was virtually no participation of women. You got this sort of female hostess and then all
guys. So clearly that’s not totally true. But you can see that when we look at nationality,
you know, most of the women are French and two of them are Polish. And a lot of those
women are very high ranking. And the two Polish women are actually members of the Polish court
who are in France because- uhm. You know coming along with the Polish King. So it’s- the types
of women that are- The way that gender and nationality interact requires to dig into
the other, the other networks to understand. So for social networks you know you could
see the females, the 17 percent, those are mostly within elite and aristocracy. It’s
something that we saw when also looking at Voltaire’s correspondence. Most of the time
when Voltaire is writing to a woman, it’s a very elite woman, a very aristocratic woman
and it’s usually asking for money. [Audience Laughs] Yea, you can see like the military
tends to overlap with the aristocracy, that’s not really any surprise because that’s kind
of their mythic origin. People who are in the government are not very often aristocrats.
People who are clergy are often aristocrats, there’s a lot of overlap. But one of our biggest
findings, is that really across all of these elites salons, the literary people are really
dominant. And this is really a huge contribution to Enlightenment studies because we tend to
think of the Enlightenment as philosophical and scientific, but when you really look at
who is attending these salons, it’s very heavily literary. And the sciences are extremely underrepresented
when we’re looking at the networks of scientists, so people like Dalum Bear. He has very, very
few scientists that he is interacting with. Five or three letters or inners. So one of
the things that we’re trying to do is to leverage linked out of resources a lot more. One of
the main goals is to be able to look at publications that are much more detailed way. And so we’re
looking, we’ve already been using VIAF, the Virtual International Authority File. And
that’s a unique identifier that’s being assigned to authors. It’s a very quick way of seeing
whether someone isn’t an author. It’s also getting closer to that Enlightenment project
that people like Blymets and Ero had of having a universal language where everything in the
world has this same name. So, VIAF has a very handy site. You can easily, you know, find
the publications in all of these sort of different international libraries. But one of the main
things that it does is help you to resolve different names. You can see this is just
for one French aristocrat. The number of names that she has in different libraries. And so
in order to resolve that problem you really do need, you do need some kind of unique identifier.
Another source is the Union List of Artist Names, the Getty. So there’s the listing for
[inaudible] the French painter. And this is really interesting, you’ve go the different
roles, artists, printer, print maker. All the different names. But you also got ways
in which he’s related to other people or corporate body. So people he collaborated with, people
he worked for, people he studied under. And then also family relationships. So when you
find painters in our data this is one of the key places to look for other relationships
they might have to people. So this is the project is supported by the Humanities + Design lab at Stanford and I really have to put in a plug for the newish tool, Palladio. So it’s
something that you can use to upload your data like in a number of different formats.
It was built largely from Mapping the Republic of Letters, so there’e a lot of features that
are useful if you’re working on any kind of letters project. There’s point to point mapping,
there’s all kinds if stuff with timelines, there’s a graph view. And all sorts of you
know you can use the time span to display changes within a network. Another cool thing
it has is faceting. So if you had, for example, if you have all of your data mapped you can
go through and say, okay these are the ten sources, I just want to see this one source,
I just want to see source number two. So it’s a really useful tool that we use a lot. I
guess that’s it. [Audience Clapping] [Audience Member] I have a question about the knowledge networks and kind of how you created them. So that seems like one of the most significant
interpretive aspects of deciding if this persons literary, this persons philosophical, how
did you go about that process? Was it collaborative process? And what do you do with people, how
do you decide when someone’s right on the edge, and lastly were there any religious
networks, even if you’re not coding by Catholic, were there difference in networks? [Conroy]
Yea, so we really did debate all of the different categories. But we really wanted to have some
kind of larger ontology. So we wanted to do- use the same terms when we mean the same terms.
It’s not useful to have five different terms for Catholic, so we did want to arrive at
similar, or the same terms, but one of the reasons we limited it was France was in order
to have a set of terms for France. The next thing we’re gonna be doing is looking at England,
and then there’s gonna be a whole question, can we actually use those categories? For
religious networks, obviously much more important to England. There are religious networks,
it’s Protestant, and then Calvinist or Presbyterian or whatever. But for Catholics we exclusively
used Catholic clergy or high ranking officials. So it’s really an inductive process where
you look at the list that you have in front of you and you say, you know, how can we put
these people into groupings where they would be more likely to know one another. So that
was a question we kind of constantly came back to. Would these people be more likely
to know one another. Yea? [Audience Member] Thank you very much, for
a very interesting talk, I was excited to hear that you were using a lot of open up
data sources and connecting the various unique identifiers out there. I was wondering are
you also contributing back to the open data, are you taking the stuff up
there? [Conroy] Um yea, it’s going to, Procope is already a linked data resource, and as
we expand it, it’s gonna be available. [Audience Member] Great, thank you very much. [Audience Member] Follow up on the other question, are people mutually exclusively put in these categories?
[inaudible] [Conroy] Oh, they can be both for sure. And you know, people like the philosophers
are often in many, many, categories. So they might be an elite because they’re going to
the salons, they might be aristocratic, they might be scientist, they might be letters.
And what we really defined is people who are in many, many salons are more likely to be
in many, many academies, more likely to have many interests. And it’s not just because
they’re famous. [Audience Member] So that visualization you put with the bar graph there, counted numbers, were there people double there? [Conroy] Yea there were people double
counted. Sure yea, here first. [Audience member] So the whole idea, or one of the ideas of social networks is about transbility, one idea from another. Are you gonna track infection
of ideas from one to another? Within salons? [Conroy] Yea we don’t have any plans so far. The next step for salons if gonna be looking at publications because the big questions for salons is are
people, you know, going in and having discussions that lead to publications? Or they’ve made
it to salons because to publication? So that’s the next step. And then sort of what are the
content of the publication maybe ten years out or something. [Audience member] I kind
of have a follow-up to that question. There’s a lot of tools in social network analysis,
so one of them is these algorithms that the tech community structure, and instead of imposing
the labels that you put out into different categories of people, you could just let the
data speak for themselves, there is a community of – [Conroy] The big issue is that we really
don’t have the complete network. It’s not like analyzing your own Facebook Network because
of the nature of historical documents we really don’t have the whole network. So we’re very
concerned about presenting something that looks as if we have the whole network. And
so the reason that we went for ontology, which has the completed everything in the network,
was to deal with that, but ideally we would eventually have the entire network. [Audience Member] Just a small remark, that URL takes you to a website under construction. [Conroy] Does it? oh. [Audience Member] So I went to the HD lab at Stanford, which is one way to
get to this, but what his website is [Conroy] Oh design! Thank you, I’m sorry. Thank you, thanks. [Audience member] I just wanted to comment, a i wasn’t clear about
the thing a minute ago about letting the data speak for itself, because in here the networks
that you’re trying to do to understand the community [inaudible] [Audience member] Well when you try to say what are these groups of people,what their knowledge network is [inaudible] The data that speaks to me is a philosopher
or scientist. [Conroy] Right, so it is [Audience Member, inaudible] [Conroy] Yea, I mean, yea
ideally we would use some kind of auto detection if we had the entire network, but it is kind
of just use imposing a kind of best guest based on what we do have. Okay thanks! [Audience

local_offerevent_note October 11, 2019

account_box Matthew Anderson


Leave a Reply

Your email address will not be published. Required fields are marked *