Graphing Yale’s Theatrical History


– Thank you all so much
for coming out today. I know it’s towards the
end of the conference, so it’s been a long couple
days if you’ve been doing both. So, we’re going to be
talking about a project that is between the Robert
B. Haas Family Arts library and the Yale Digital Humanities Laboratory and it is called [email protected] So we’ll get into that in a moment, but first to introduce ourselves,
my name is Alex O’Keefe. I’m the arts digital projects librarian at the Haas Arts Library. – And my name’s Catherine DeRose. I’m the program manager of
the Digital Humanities Lab. – So, the conference is all about networks and a lot of it has been about the Yale, how we work together. But we’re looking at another
from of a Yale network, and it’s more to do with the history. So, [email protected] is a project that deals with Yale’s theater
history with Drama 37, which is an archival
collection of programs from the Yale School of Drama and the Yale Repertory
Theater, dating from 1925. And our project, they go up to 2016, so it’s quite a long amount
of time that we’re covering. And if you don’t know a lot
about the Yale theater scene, there are a lot of famous
alumn from the School of Drama, including Meryl Streep, Angela
Bassett, Sigourney Weaver, lots of them, I’m not gonna
try and name them all. So we have, quite often, questions about people who went to school here
and what plays were they in, how many times they
perform on certain stages, things like that. And so one thing that
we can’t do right now is really answer those questions because they’re all in an
archive, one program at a time, and we don’t have a list of
what everyone’s been doing and everyone’s been up to. So, we wanted to increase access to what is a minimally-processed archive by library standards, generate a database of
Yale theater history, and then also, ultimately for us, answer reference questions. So, the project is inspired, though. We can’t take full credit. (laughs) It’s inspired by an NYPL,
New York Public Library, Labs projects, just called
Ensemble, and that was in 2013. And so when ours launched,
we called it [email protected] to clarify the difference. But the thing that makes
these projects special is, if you all know about gathering texts from different documents, you can do OCR, which is machine reading, where it just gets all of the text. We didn’t want to do
that with these programs ’cause there’s a lot of text
we don’t need especially, like advertisements from
the 50’s about watches. We don’t need text from that. But also a more important thing of it is that we wanted to find
semantic relationships between person and role. So we wanna know if Meryl
Streep was in a play, who did she play. Those sorts of relationships
are really important and in terms of the networks
we’ll talk about too. And it’s also hard for computers
to see those relationships and human judgment is
extraordinarily important for it with this sort of format and human understanding
of plays and programs. So the project itself has been going on for quite some time now. It had a lot of background
tasks that had to happen before it launched, so
they digitized the programs that I was talking about in the archives with student workers’ help
in the digital collections at the Arts Library. Determining the organization
and presentation of the site, all of those sorts of baseline
things you have to decide. And then we use a platform called Scribe, which we wanted to customize
workflows for theatrical data. So Scribe is to mark and transcribe things in a crowd-sourced way, which
is what our project does, but there is a lot of
customizations that happened before it launched. We learned a lot of lessons (laughs) from that process, as one does. So we had usability
testing before going live, which did happen, but it really helps to make sure that’s a priority. Fixing errors and preventing junk data like that testing of the platform itself. And then how to decide when marking and transcribing is complete. It’s kind of a workflow thing we had to determine in the early stages. So it launched in May of 2017 and this is, I’ll actually navigate to it in a second. It’s a little hard with the dual-screens. But it launched and I’ll show
you a demo of how it works. No I won’t, what did I do? Is it this one? – No, it’s not. – Oh, okay. So, let’s try getting out of this. – You have to exit fullscreen
to be able to get that. – Sorry? – If you exit fullscreen.
– Oh yeah, there it is. Boop boop boop. Okay, sorry everybody, please
bear with me for one moment. There we go, okay. So if you just go to Ensemble, which hopefully I’m typing right. Can you see it Cathy? – There’s an error. – God, I knew it! (laughs) – After the N, yeah. Oop, there it is, it pre-loaded. – Okay, I was like I
know you have it on here. Okay cool, so this is the
site live and in-person. So to get started, you log in. We have a lot of different
options on how to do that, including your netID if
you’re a Yale person. Do you mind if I log
in under your account? (laughs) And then you have two options here, which is marking and transcribing. So the mark workflow is that over here we have all the
different things we’re looking for and it’s really neat, you
can see all of the program one page at a time if you’d like. You can see those fun
advertisements I was talking about. But then we’re looking
for pages like this, which have the title and
other people involved in it. So you can just draw a box
around the thing and say “Done.” And then you can keep
doing that to completion or if you’re interested in transcribing, this is how transcribing looks. And we have custom boxes
for each type of thing, so if it was a person, it
would ask you for their role as well as their name. In this case, it’s just the
title and possible subtitle. So what an appropriate one, Let’s Do It. I did not plan that. I’m very happy about
the serendipity of it. And then you can say basically
“I’m done transcribing “and I wanna go back.” So it’s a really cool platform
and through doing this we have done a lot of programming
to get people involved. If I can get back to
the presentation, great. So this is in case the website
was down for some reason and it wasn’t, hooray! Oh okay, so we’ve had a
lot of really fun events that we call transcribathons
and transcribe challenges, where we’ve had folks
come out into our space in the DHLab space in order to learn how to use the platform
and contribute to the project. It’s been really fun. We also had a visit to Arts
Library’s special collections, so you could see those
programs you’ve been working on live and in-person, which
was really exciting. And then we have trivia
rounds at every single event, where we also do Yale Rep
and Yale School of Drama specialized trivia, written
by our expert drama librarian. And so, yeah. So we have a lot of ways that
we ask people to get involved and in so doing, we’ve
gathered quite a lot of data. If you might have noticed, we have the platform
organized into six eras. We have finished the first era. It’s extraordinarily exciting. (laughs) One of six is done, that’s
over 200 programs of data. And so now we have to think about what we’re doing with that. So I am in the process of cleaning it and we have steps to get through it all. So initial cleaning using OpenRefine. Removing duplicate values
because multiple people can transcribe the same program, in Excel. Checking the program, we right now are training student workers
to help with this step because we are trying to make
the cleanest dataset possible so there’s a lot of steps
in this to make sure that essentially any errors
are caught along the way. And then normalization
to improve searchability, because ultimately we wanna
upload this data into Findit and then we also want to
have an interoperable dataset that people can access. We’re still determining
what that’s gonna look like and how much linked open data
practice we can or can’t use. So yeah, so we’re normalizing names using, right now we’re just kinda homebrewing it close to the Library
of Congress standards, but we haven’t had a chance
to start cross-referencing. And then roles, we’re
using Australia’s stage has a set of terms to use. And so we’re normalizing
everything so that researchers could say, in
case they’re interested in “I wanna know everyone who
ever did X staff role,” you could look that up, kind of thing. Because the fun thing
with historical programs is semantics change over time, so when people’s titles change over time. As you can imagine from 1925, those titles are not the same as 2018, so these are the sorts
of steps we’re taking to try and maximize our data
for researchers ultimately because that’s what this project is for. And then we have a final check process where we check the data against itself and make sure everything looks good. And now I’m going to
turn it over to Cathy. – Thank you, Alex. So this is an interface
of one of the tools that Alex has been using
and I’ve been using when we’re thinking
about cleaning the data. It’s called OpenRefine,
it’s free, it’s open source. What’s really nice is it’ll let you, it’ll ingest a spreadsheet
that you wanna give it and it’ll give you ways to filter down and actually see specific categories in kind of clusters. So for an example of
that, this is one where I’ve asked it to look at
our People Names column and see if there were some
that it thought should be clustered together, that
are currently separated. So this is where it’s
catching a lot of that name normalization that
Alex was mentioning where, yeah these names look identical, it’s just that somebody
put a comma before Jr. when they were transcribing
it on one program and in the other program,
there wasn’t a comma. And so to the computer, those
are actually different things, because that missing
comma actually matters. But to us, we wanna treat
those as the same person that we know they are. And so this is the stuff
that Alex has been doing a lot of the work towards, reconciling these slight
differences in names and having to identify at one point an actor might
use their middle initial, and the next year they
might decide not to. And so having to figure out
are these the same people so that way, we’re not
double counting them when we actually go to visualize
them as a network graph. So that’s one part of the cleanup. The other part of the cleanup
is actually getting the data into a format that’s amenable
to creating a network graph. So for just a little terminology, for network graphs, networks are studies, they’re visualizations of the
relationships between things and how those things are connected. Things in this case, for
the Ensemble network, are gonna be people and the
thing that’s connecting them is whether or not they were in
the same performance together in some way, whether that
was as a staff member, or as an actor or as the director. And so while we have all that information in these spreadsheets, in order to create it as a network graph, we actually need to present it
in two separate spreadsheets. We need to have one
that’s called a node table and nodes are your things, so in this case it’s going
to be our people category. And so we’ll have a unique ID column, so that we don’t lose
all that wonderful work that Alex is doing for
the name normalization. If there were two Meryl Streeps, if another one comes around, we wanna be able to differentiate them. And in addition to that and
in addition to their name, we wanna have what their role was. Were they a cast member,
were they a director? We wanna capture that information. So we have that in our node list. And then we also have
what’s called an edge list or an edge table. And the edge is the word that people use to describe
relationships in a network. Or depending what field
you’re coming from, you might hear link or tie is another phrase that people use. And so in this case our edge is gonna be two people that are
connected to one another and the play that’s connecting them and the year that that play took place. So that way, if we have
repeat performances, we can clearly separate those out in the network raft that
we’re gonna produce. And so these are just a
couple snapshots of some Python scripts we wrote to
help automate some of that work of the network creation. And what we actually get is
something that looks like this when we run it. This is your classic
hairball example in networks. Alex and I think it looks a
bit like a Rorschach test. – Yeah.
– So I see like a lung and a heart. Alex has something very
different. (laughing) But in either case,
what we’re not seeing is a really nuanced pattern that gives us a sense for
what’s happening in the network. And that’s partly because it’s so large. So when even this is just a subset that we’re working with right now, and it’s already 991 nodes. So there are 991 people in this network. – And it’s 52 plays. – 52 plays.
– So this is only 52 plays (laughing)
when I was saying there’s over 900 programs total
going to happen eventually. So this is a small portion. – And the edges, the connections
of people in these plays are over 23,000. And so seeing it all in mass, it’s kind of hard to figure out
more than a general sense of there seem to be some large clusters. But we need to actually delve
into it and experiment with it a little bit to tease out
some more meaningful patterns. And so what I’m gonna show
now are a few different ways that we’ve started teasing that out. So, one of the things we
can adjust is the layout. With network graphs,
layout is semi-arbitrary. You can move a node, which
is represented as a circle, a person here, into any place
on the map that you want. So what doesn’t matter is if two nodes look like they’re close to
each other on the layout, what matters is that there’s that line, the edge connecting them. But since networks and
since layout is flexible, it means we can manipulate
it a few different ways and see if we can find
any patterns that emerge. So, what we’ve run with these examples are two different
force-directed algorithms and essentially what they’re both doing is they’re turning the network
into a magnetic field so that the nodes repel one another but the edges act like springs that try and keep things
kind of close together. And the reason they have
different layouts is because the way, when they transform
it into a magnetic field, they score things slightly different and that’s helpful in our purposes because they point out different patterns. They’re both statistically valid ways of representing the network, but they’re useful in different ways. So for the Force Atlas 2 algorithm, it’s really helpful for finding what are called brokers in the network, so the people who connect otherwise disconnected
subgroups in the network. So these are the people like, you might see this orange circle here is connecting this group
and that group together. It’s also connecting this group, and so by stretching it out, we’re able to see who are those people who
are really functioning as the connector from
otherwise disparate groups. Whereas the Fruchterman Reingold layout is doing a nice job for showing some of that clustering in a little more detail, where we can see where the
really dense clusters are, there’s a lot of people
starring in something together, and then we can see peripheral connections of people who might’ve just
been in one performance with that otherwise really
densely connected group. So part of our work over the
next few months is gonna be really diving into these relationships. And in addition to layout, another way that we can work with this is through affecting color. So, we’ve rendered it in two
different ways right now. The one on the left is
using that information from our node list that we captured of whether somebody was a
cast member, a staff member, a director, or a playwright. And the category Multiple there
means that they have been in multiple plays where they’ve had an equal number of times
that they were a cast member and equal number of times
that they were a staff member, so we really couldn’t categorize
them as one or the other, since they really played both equally. And there’s ways that we
might refine our categories moving forward, but for right now that’s one way that we’ve rendered how people were involved in the plays. And that gives you a nice glance of well yeah, we can see
that there are a lot more cast members than
others, not surprisingly. Staff members are also
really densely populated. And one of the things that
was interesting to us too, looking at that one on the left as well as kind of in
the Force Atlas 2 one is that we were surprised at the role directors were playing. When we were originally talking
about what we might see, we were thinking that it
would be staff members who would be the people
connecting the network because staff members would go on to be involved in many shows. But it turns out that
it was returning actors were really important for connecting performances across time, whereas staff maybe have a deeper role in creating connections for a
more condensed time period. So in the 50’s, it might
have been staff members who were really making the connections, but if you want connections
between the 1950 plays and the 1940 plays, it might be the director who’s actually making the connection. And then the other way that
we’ve experimented with color is with modularity class,
so that is an algorithm, it’s a community detection algorithm, so it’s trying to find
subgroupings within the network. And so as Alex said,
we have over 50 plays. Just over 50 programs
that are part of this, but what the computer
found is 10 groupings. So the computer thinks there’s
only 10 groups in here. And so that suggests to us that again, we have people who are
in repeat performances to such a degree that they have formed kind of their own sub-network. The way that modularity class works is it looks to see that
there are more connections within the group than without the group. And this is an algorithm, all these have been generated
in a program called Gephi, which is also free open-source software for creating networks. Another experiment that
we’ve run is with size. So these are two metrics
that are trying to capture a node’s importance and they’re doing it in
slightly different ways. So betweenness centrality is
looking for how often is a node on the shortest path to any
other node in the network, so that tends to be somebody who’s really connected with a diverse group. So a good example of betweenness
centrality for travel is Chicago’s O’Hare airport has
a high betweenness centrality for travel in the US. If you’re trying to
get across the country, you might have to go through O’Hare. You certainly have to go through O’Hare more than you might have
to go through somewhere in Florida, for instance, if you’re trying to
get across the country. So that’s roughly how
betweenness centrality’s working. But one of the things that
we’re gonna have to think about is betweenness centrality
doesn’t always work well at scale because it’ll start to
distort in various ways, versus something like
eigenvector centrality, which is also trying to
measure a node’s importance. But it’s more the “it
crowd” kind of measure, where it’s looking at this person is connected to other people
who are really connected. Versus this person’s connected to somebody who only has one other connection. And so the bigger a node
is, the more in theory, according to you know the algorithm, the more influence that person has in communication in the network. And so what we can see by comparing these is we see slightly different patterns. So whoever this person is in blue still has a decent betweenness centrality, but really it’s the person in this seafoam-y green color here who has a higher betweenness centrality but they’re not very important according to the eigenvector centrality. And all of this is,
we’re humanities people. We’re not gonna rely on
the statistics as like this is an answer to something. But what it is, is it’s a
prompting to us to say like, Well, who are these people? Can we go back and look
at their connections? The computer thinks there’s
something interesting there, do we think there’s
something interesting there? And so it’s more of a visual cue for us about how we might dive
into this otherwise really large and kind
of unmanageable network. And then two more ways that
we’ve experimented with it. One is across time, so we wanted to see how do the connections change
over the period that we have as our subset, which
is 1925 through 1952. And so what this is showing us, again it’s just those
connections forming over time, but it suggests that the modularity color scheme that we were using is actually really good at telling us when people were performing. It seems to be capturing decade
or sub-decade information. Like there’s your 1950 cluster over there, our 1920 cluster over there, starting to move into later 1920’s, 1930’s at the bottom here, 1940’s, earlier 1940’s, and then 1950’s. And why that is useful is that, again, gives us some different
ways of thinking about, well the 1920’s, that blue
and black cluster over there, that’s massive and it’s
really densely connected in a way that the 1950’s isn’t. And so does that mean were the programs different back then? Were they just smaller casts in the 1950’s compared to the 1920’s? Were there just people in
repeat performances more? Was that a custom in the 1920’s that drops off in the 1950’s? Again, it doesn’t give us an answer, but it gives us a way to
ask questions of the dataset that we can go back to and use our traditional humanities
methods to delve into more. And the last thing that
we’ve been experimenting with is partitioning the network, and so that’s pulling out specific subsets that we wanna look at more closely. And so I’ve put up just
five of the ones here. So we have all of our
cast members over there, directors in the middle, and then staff. And then I pulled out two
of the modularity groups that according to the
eigenvector centrality seemed really important. And what’s useful here is again, it gives us a sense of casts
are really densely connected, but particularly in this cluster, like why are these casts not
quite as densely connected? We might go and look at
who are those people, what performances were they in, that time. Are we seeing the time difference? Are these all the 1950’s cast
members and these are 1920’s? For directors, what we loved
about it is we’re like, oh, we didn’t expect to see directors really connected to each other at all. But it turns that sometimes
directors would return and actually star as a cast member in a fellow director’s performance. And so this showed something to us that we wouldn’t have thought
to ask of the dataset, like are directors ever cast members? They were. And then again, staff we can see, we can start to look at
how they’re connected and how they might differ from the connections of staff members. And similarly for modularity scores, again it just gives us a way
of delving into these subgroups which we now know thanks
to the time slider seem to be roughly correlated
with the performance date. And then for the last
slide that I’ll show is this one that actually
answers the question that Alex had started with. So, one of the original
aims for the project was Lindsay King, the associate
director in the Arts Library, often gets the question, “I wanna know what this person starred in. “I wanna know who they starred with. “I wanna know what their role was. “I wanna know their years at Yale.” All these things that, or answers that we didn’t
have to hand already. Now we can start to
answer those questions, both from the spreadsheet but
also in an interactive way where we can see, so Sarah Sigler. She’s light blue because she’s
in that Multiple category, meaning, in this case, she was both a staff
member in a play in 1946 and then she was a cast
member in a play in 1947. We can see what those plays were. We can see who else was
with her in those plays. If we discovered somebody
of interest, like if the person who’s like, “I
wanna know about Sarah Sigler.” also wants to know about
whoever this person is, they can now see their
network very quickly as well without having to go back to the archive and do a ton more research
to tease out these patterns. They’re now more easily accessible for people to play around with. And so that’s very exciting
to us because it suggests that this is gonna be a good way moving forward once we get all of the data in, but we’re still gonna have to tease out, if it’s already that dense of a network with just the subset
that we’re working with, how are we gonna scale it up? To be able to visualize it more. And so this is just our thank you slide, but then we also have,
if you’re interested, we also could show you sort
of the live Gephi network. We have that running on the machine. But otherwise, we’re
happy to take questions. Thank you so much. – [Audience Member] So I’m
doing it while we’re talking. – Oh, that’s fantastic! Thank you.
– It’s a lot of fun! – Yes yes yes yes. – It’s fun!
– That’s what I wanna hear! Okay. (laughing) – It can be really therapeutic
too, just kind of like, to go one by one.
– Yeah! I just, you know, we’re a senior talking– I mean, look at this. (laughing)
It’s quite a– It’s a little addictive! – Yes!
– Yes. – Good.
– That’s the goal. (laughs) – [Audience Member] No really,
I didn’t pay attention to– (laughing) – Yeah, that’s great to hear.
– And then we have like I said, those really fun
events called transcribathons. We’re trying to introduce
some new stuff in the fall, so if you wanna get on our email list, just let me know and we
send you info about it. But it’s opportunities for
people to come together and do this work together because as fun as it is, it’s working isolated, it just kind of gets old. And so we all come
together and we’re like, “Oh my god, listen to this name. “This is a weird name. “Oh my god, this play, I never thought this
play would be happening.” So yeah, so it’s really fun– – [Audience Member] Well, my daughter is actually starting theater at UCA right now – Cool!
– Oh, fantastic. – [Audience Member] And I’m thinking this would be really cool to do with her! – Mhm.
– Yeah, this is a great– – Well it’s a longer term goal, but something that’s on our mind too is now once we have this network created, there are records of Broadway plays and where people have starred and can we see some connections
if we sync our data up with Broadway’s dataset
and starting to see who went on from Yale to do other things. – So we’re trying to set some precedent with how we’re handling our
product and approaching it, but also how we’re cleaning the data in the hopes that other
people will take on the mantle with their archives or collections and start doing similar work. And then the more we do it,
and the less siloed it is, the more opportunities there are to explore these connections. – Yeah.
– Yeah. – [Audience Member] Oh, that’s great. – Yeah.
– Oh, it looks awesome. – It’s fun. (laughing) – Okay, we’re so thrilled to
hear you like it. (laughing) That’s great. – [Audience Member] But can I now, so if I’ve only done part of
it and if I need to leave, will somebody pick up where I left off? – Yes, currently.
– So it’s while we’re on Scribe, I was gonna say. So we might be switching engines. We’re working on it this summer, but right now the answer is yes. So you can come and go as
you please, essentially. – [Audience Member] Oh, okay. – In our newer one, it’s gonna
be a little bit different of a workflow and it will be you have to do something to completion, but it’ll be clear that that’s
how it works, essentially. So when that change
happens, you’ll know it if you’ve been playing on
the platform this whole time. But for now, yeah you can
totally drop in and out as much as you feel like. – And one of the great
things, one of the reasons too that we’re switching to the new platform is to actually gonna let you see all of the things that
you’ve contributed to, so you’ll see like, “I’ve
transcribed 30 programs.” or something.
– Yeah. And it’s really fun too ’cause you can make your
own custom collection. So if you found something you really like, you can favorite it essentially
and you can explore it and then we’ll also have a chat board So there’s a lot of new features that will get people able
to engage in new ways. – [Audience Member] So
when I come back on, this unfinished playbill is
what I’m gonna be working on? – Not necessarily.
– Or will someone else pick it up?
– Yeah, if you’re doing just
the mark and transcribe top key buttons,
– Yeah. – It’ll just dump you into something the program has decided is important. So you may or may not
have the same one twice. And so there’s one way to navigate on the Era page, where you have
a little bit more control if you really care to do a
program all the way through. So let’s see, if you
go back to the homepage and scroll down, so these are
the eras I was talking about. This is the one that’s done now, so you can’t transcribe
anything in that era, ’cause it’s done, which
is, again, very exciting. (laughs) – Oh, that’s cool. – Yeah, so if you click on any of them, but suit your fancy. You’ll see all the
playbills from that era, which is really cool and then
it has the mark and transcribe when you hover over any playbill. The thing I will say is it moves them almost every time you get back on, so that’s where the findability
can be a little difficult if you’re trying to do repeat
things, but if you are like, “I really do just wanna
work on something specific,” you can go through this
way on this platform. And again later we’ll be changing that. It’ll be totally random,
there won’t be that control. But for now, you’ll have this ability to kind of pick and choose. – And even when we switch over to the new platform
for the transcriptions, this will sort of become an archival site where you’ll still be able
to browse the programs if you wanted to do that too. – Yeah, so until they’re in Findit, we don’t wanna deprive people
of this cool tool we’ve made to explore, and so what’ll happen is on the new platform, there’ll
be metadata that you could filter this to find something on here, if you really just want a
view-only version of it. Or again, you can favorite
it on the platform and also like stuff, so there’s a lot of cool
features on the way. (laughs) – Cool.
– Yeah. – Yeah, this is neat. – It’s a fun project. It’s also really fun just to look at how things have changed
over time and since I am not even neck deep, I’m like this deep in
the data at this point. It’s fun in that I made a lot
of assumptions about things from looking at names over and over again and then as soon as we
made the network chart, Cathy knows, ’cause I
was like, “Wait, what? “That, I didn’t expect
that at all, wait what?” And it’s just this really cool way to revisualize and rethink about it, even after I’ve been looking at it literally for weeks, right? (laughing) So, yeah. And then I think we should show them the – Sure,
– Gephi thing, it’s pretty cool.
– Yeah. – So all of those screengrabs
Cathy was showing you, on Gephi if you ever get
the chance to play with it or use it or are interested
in network graphing, it’s a lot more interactive. So all those things
Cathy was talking about where she can make changes
or move things around, this is what it would
look like in the program. So anytime you hover, you
see the name and the plays and you can zoom in. Make those algorithmic changes,
like Cathy was talking about so it’s really fun to play with. (laughs) And we’re hoping eventually, we haven’t figured out what it’s gonna look like, but some sort of public data visualization for people to have these
sorts of experiences with. The DHLab is trying to, we’re trying to think through
ways to make that work, but for now, any individual researcher once the data’s available
can do this themselves. Possibly with some training
of course, but still, yeah. So this is a really cool
way to think about it and look at it. – And what’s nice is, so a lot of the algorithms
that I was showing, those are things that Gephi provides in the statistics pane here, but like role majority, that was something we
had in our spreadsheet. So any additional information we capture about the performances or about the people are things that we can then filter by or color code by later on. So, lots of options. – Yes, and we’ve, again,
tried to make the data in such a way that researchers have those options available. And we also are trying to keep the original entries,
because we ask people to type exactly what they see, so if you were interested
in the different ways Frank McMullen spelled his name, you would have a way to find that out. Or if you were interested in
how titles of casts and staff have evolved over time, you
would be able to explore that. So if you were like when
did production manager become stage manager, or
these sorts of connections. Dramaturgs who are drama historians are more interested in those things, but we’re trying to
think about our audience and who will be using it. So it’s really fun! – [Audience Member] Did you guys organize your eras based on the directors? And if so, what was the
reasoning behind sort of naming the eras after the person? – So, I wasn’t there for
the beginning organization, I should say before I start talking, but my understanding is that these things are all in our
archive just chronologically, so it really wasn’t subsets. And thinking about potential users and people interested in this, a lot of thought went into alumni and how they might wanna
see things they were in or their friends were in and work on those programs specifically. So they were trying to think
of a way to organize them in smaller subsets and I know
they landed on this option after a lot of different
talk and discussion of ways that could happen. So instead of making
it arbitrary and saying every 25 year chunk,
or something like that, they wanted to have some
sort of meaning as to, until 1966 though, there’s not
a name because these two eras don’t have the same– Again, you can think about the
mystery in how things change and how we think about programs within a larger theater network over time. – [Audience Member] So
for the Founding Era and the Department of Drama, there obviously was someone at
the helm of the organization, – Right.
– It’s just obviously a lot of things changed, or they (mumbles) for
other schools, whatever. So how did you guys name
like Department of Drama, Founding Era, since those
are kind of separate from directors’ names? – So that’s where we do
have some sub-information on each era to essentially explain it. So this one is, so Department of Drama essentially existed
before Yale Rep existed and so that’s where Department
of Drama is by itself, because this is when that started. And then when Yale Rep starts, is– I can’t remember what Founding Era, ’cause I am not a drama historian. Yeah, so this is when it starts having a separate master of fine art– they change in some way and then Yale Rep with
their directors comes in and affects the rest of the groups. – Cool.
– Yeah. Yeah, thanks for your question. And if I ever don’t answer
something to satisfaction, (laughs) just let me know. I’m happy to follow up
with more robust responses. – Yeah, as we said, as hopefully was clear in our last slide, we have a big team that’s working on this. – And a big team that has
been working on it over time, so I’m newer to the project in the large scheme of our time table. – [Audience Member] (mumbles) – Good!
– That’s great, yes! – I’m so happy to hear it. – Thank you so much! We’re happy to follow up with anyone too over email or things, if
you have any questions.

Leave a Reply

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