Teaching on the Intersection of Visualization and Digital Humanities

Stefan J

anicke

University of Southern Denmark, Odense, Denmark

Keywords:

Teaching Visualization, Digital Humanities, Journalism.

Abstract:

Visualization as a means to generate hypotheses and to communicate insights on digitized cultural heritage

data sets has become more and more important in the recent years. While many digital humanities researchers

transform their data in order to be processed with ready-to-use tools, others engage in interdisciplinary col-

laborations with visualization scholars aiming to design novel solutions and interactive visual interfaces as

occurring data features are more carefully mapped to visual attributes. Many of those collaborations suffer

from loss of time at the beginning of the project as scholars with diverse research backgrounds require to

understand each others’ mindsets, ways of thinking and research interests. This paper reports on three years

of teaching visualization design for digital humanities projects. It provides an overview of theoretical course

contents, practical training, collaboration setups—involving computer science, digital humanities as well as

humanities students who experienced typical collaboration obstacles—and remarkable project results.

1 INTRODUCTION

For three years, I taught a course on information visu-

alization for the digital humanities equivalent to ﬁve

ECTS credits. Due to varying student groups through-

out the years involving Masters students of the sub-

jects computer science, digital humanities, humani-

ties and journalism, the course included a balanced

training of technical, domain-speciﬁc and interdisci-

plinary aspects of digital humanities research involv-

ing visualization for hypothesis veriﬁcation and gen-

eration. In this paper, I want to share my experiences

in teaching theoretical and practical contents of the

course that may serve other teachers as guidelines in

designing related courses on the intersection of visu-

alization and digital humanities.

My course included a theoretical training dis-

cussed in Section 3 in the form of a lecture giving a

broad overview of general visualization aspects, dig-

ital humanities as an interdisciplinary domain, means

to improve close reading in a digital environment as

well as typical distant reading techniques. The the-

oretical training was complemented with a practical

training discussed in Section 4, in which the students

carried out little visualization projects themselves. On

the basis of related research questions and data sets,

the students processed the data according to visual-

ization ideas and developed Web-based applications

used for interactive visual data analyses. The main fo-

cus is that participating students learn the challenges

of interdisciplinary work, and, independent on their

studied subjects, gain an understanding on research-

related topics and strategies how to design and imple-

ment visualizations that appropriately reﬂect data fea-

tures visually, serving to support related data analysis

tasks.

2 RELATED WORKS

The main goal of my course is to prepare students

for interdisciplinary, research-related project work.

To achieve this goal, the course structure is mainly

inspired by Healey’s research-based teaching ap-

proach (Healey, 2005), who outlines different ways

of connecting research and teaching that are reﬂected

in different carried out activities in my course. Ac-

cording to Dohn and Bolin (Dohn and Dolin, 2015),

research-based teaching is the best suited approach

as (1) the content and form of teaching is largely

focused on current research endeavors in digital hu-

manities, (2) my own background is on the intersec-

tion of visualization and digital humanities, the course

is designed on the pillars of the visual text analysis

process (J

anicke et al., 2017) and includes my own

research outcomes, and (3) students actively partici-

pate in current research, e.g., by carrying out projects

driven by real-world research inquiries.

100

Jänicke, S.

Teaching on the Intersection of Visualization and Digital Humanities.

DOI: 10.5220/0008987101000109

In Proceedings of the 15th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2020) - Volume 3: IVAPP, pages

100-109

ISBN: 978-989-758-402-2; ISSN: 2184-4321

To serve the students with a fundamental theoret-

ical basis of visualization, I drew suggestions from

related works in the visualization ﬁeld too (Dykes

et al., 2010). The course structure is in line with Mun-

zner’s subdivions “core readings”—complying to my

theoretical contents section—at the beginning, and

“individual student presentations”—complying to my

practical training section—to the end of the course.

The advantage of students having done “the bulk of

the readings before they must choose project topics”

is especially reﬂected in the high quality of student

project outcomes. As opposed to preferred litera-

ture (Spence, 2007; Ware, 2012) used by other teach-

ers in the ﬁeld (Kerren et al., 2008), I assess the rather

application-driven approach to teaching visualization

design principles (Munzner, 2014) as best suited for

my interdisciplinary course.

3 THEORETICAL CONTENTS

The main objective is that students learn all relevant

aspects of interdisciplinary digital humanities projects

that aim to include visual interfaces as valuable in-

struments for humanities scholars to gain insights into

their data, most often being text. Moreover, students

should aquire the capability of implementing digi-

tal humanities projects, and to discuss project direc-

tions with—dependent on their own backgrounds—

humanities or visualization scholars. The contents of

my course are related to the visual text analysis pro-

cess (J

anicke et al., 2017) as illustrated in Figure 1.

Most importantly, projects should start with a central

idea, the text analysis task that inﬂuences all further

steps throughout the project. Throughout the course,

most related works were discussed in the context of

the visual text analysis process, but it is important to

note that not all of them were based on raw textual

data, i.e., some projects based their research on re-

lational databases containing structured textual infor-

mation. The text analysis task is typically directed

towards a text source that is of interest for the human-

ities scholar. Data transformation and the developed

visualizations strongly relate to this task and the un-

derlying data. After a broad overview of the visual

text analysis process and exemplary projects, the fol-

lowing contents composed the theoretical training of

the course:

• Visualization Design: First, I gave an overview

of general visualization guidelines based on

Munzner’s book “Visualization Analysis and

Design” (Munzner, 2014). Other visualization

text books can be used as a basis for teaching

related courses, e.g., Colin Ware’s “Information

Visual Text Analysis

close reading

humanities scholar

text analysis

task

text sources

insight

data transformation

distant reading

Figure 1: Visual text analysis process (J

anicke et al., 2017).

Visualization” (Ware, 2012) choosing a rather

theoretical approach or Cao and Cui’s “Intro-

duction to Text Visualization” (Cao and Cui,

2016) for a mere application-driven approach.

However, Munzner’s practical approach of

breaking down the visual design process into

the questions WHAT (data is to be visualized),

WHY (a visualization is necessary, in other

words, what are the user tasks to be supported),

and HOW (data can be visualized to support

the intended user tasks) provides a well-suited,

comprehensive overview of the major aspects

related for the design process of visualizations

in the context of digital humanities projects.

The great opportunity is that students learn,

independent on their academic backgrounds

in computer science, digital humanities, social

science or humanities, the same terminology

for visual design eabling them to purposefully

discuss project ideas. In addition, conveying data

and task abstraction strategies is very important

to shed a light onto particularities and commonal-

ities among the many concrete projects discussed

throughout the course. Finally, one major reason

for teaching on the basis of Munzner’s works

is the nested model (Munzner, 2009). My own

experiences (J

anicke et al., 2016; J

anicke and

Wrisley, 2017; Khulusi et al., 2019) outline that

carrying out digital humanities projects based on

the nested model model signiﬁcantly increases

the likeliness that the developed tool serves the

intended purpose because domain experts with

their knowledge in the targeted research ﬁeld are

strongly involved in the development process.

The interdisciplinary exchange necessary for

successful digital humanities research students

experience in the practical training.

Teaching on the Intersection of Visualization and Digital Humanities

101

• Digital Humanities: Now that all students were

supplied with a basic knowledge of visual design,

we discussed the digital humanities as the focused

domain. This included major historical develop-

ments in the ﬁeld, e.g., Wladimir Jakowlewitsch

Propp’s “Morphology of the Folktale” published

in 1928, in which he offers the ﬁrst known quan-

titative view on a collection of Russian folk-

tales (Propp, 2010), or Roberto Busa’s “Index

Thomisticus” project that started in 1948 with the

lemmatization of Thomas Aquinas’ works (Busa,

1980). The overview ended with a discus-

sion of recent driving forces including Franco

Moretti’s “Graphs, Maps, Trees” (Moretti, 2005)

who controversially promotes visual representa-

tions for texts, the Culturomics project (Michel

et al., 2011) being one of the ﬁrst attempts to

quantitatively analyze digitized text collections

in the ﬁeld, and Matthew Jocker’s “Macroanaly-

sis” (Jockers, 2013) giving an overview of the sta-

tistical fundaments for such analyses.

• Close Reading: Munzner asks for mutual ex-

change of domain knowledge and research ap-

proaches in interdisciplinary projects (Munzner,

2009). As most students of the course are not fa-

miliar with the close reading technique (Burke,

2012; Mesmer and Rose-McCully, 2018) being

a fundamental method in traditional humanities

scholarship, one lecture was dedicated to it, and

the students could perform a close reading them-

selves in order to reﬂect on the main idea of a text,

to think about narrative style, vocabulary, syntax

and the context in which a text was written. Con-

textualizing our own work (Cheema et al., 2016),

I conveyed the capabilities of how visualization

can support digital close reading (Kehoe and Gee,

2013) ranging from manual annotations to quanti-

tative distant reading statistics.

• Information Seeking Mantra: On the bridge be-

tween close and distant reading, I discussed the

main differences among both techniques using

Shneiderman’s mantra “Overview ﬁrst, zoom and

ﬁlter, details-on-demand” (Shneiderman, 1996).

While distant reading or macroanalysis are the

terms used by digital humanities scholars for

overview, close reading or mircoanalyis relate to

details on demand. For the continuum in between,

zooming and meso reading (J

anicke and Wrisley,

2017) are the established terms. Many visualiza-

tions that have been designed implement the infor-

mation seeking mantra as the humanities scholars

are provided with a distant reading, and means of

interaction lead them to interesting portions of the

underlying data source.

• Distant Reading: Throughout the course,

I discussed the main related distant reading

techniques—heat maps, tag clouds, maps, time-

lines and graphs—with the students both in the-

ory and practice. On the one hand, concep-

tual and technical details are outlined such as for

tag clouds (Sinclair and Cardew-Hall, 2008; Vie-

gas and Wattenberg, 2008) and timelines (Aigner

et al., 2011; Bach et al., 2015; Brehmer et al.,

2017), on the other hand, practical examples

from related works underpin the value of distant

reading techniques for (digital) humanities schol-

ars (J

anicke et al., 2017). Related works were

chosen based on their actual applicability to dig-

ital humanities research. For example, Collins’

Parallel Tag Clouds (Collins et al., 2009) that

have been applied in the Trading Consequences

project (Hinrichs et al., 2015), was discussed.

This way students get to know strategies how

to adapt existing techniques to support research

inquiries in digital humanities. Further, basics

of drawing graphs (Tamassia, 2007)—being the

most frequently occurring data structure in dig-

ital humanities research—were conveyed. My

motivation to stress this topic is the rather low

quality of graph visualization design in related

works. Digital humanities researchers typically

apply ready-to-use tools (Bastian et al., 2009)

to generate visual output quickly, disregarding

means to appropriately map graph-structural to vi-

sual features. The beneﬁts of careful graph de-

sign were dicussed on the basis of our work on

TRAViz (J

anicke et al., 2015), a graph visualiza-

tion reﬂecting the features of aligned text editions

on sentence level, which was compared to previ-

ous approaches (Haentjens Dekker et al., 2014;

Andrews and Mac

e, 2013) using a standard graph

drawing library. Finally, as uncertain information

is often comprised in digital humanities data due

to the long history the data refers to, current re-

search endeavors to tackle this problem (B

orner

et al., 2019) were discussed in the course.

4 PRACTICAL TRAINING

One of the most important driving forces for the

practical training was my own interdisciplinary back-

ground. When I started working in digital humani-

ties projects as a computer scientist (J

anicke, 2016),

me as well as my collaboration partners from the hu-

manities needed to learn speaking the same language

and to understand each others’ research interests and

ways of thinking. I admired that students get this ex-

IVAPP 2020 - 11th International Conference on Information Visualization Theory and Applications

102

Table 1: Projects carried out in three years involving students and researchers with backgrounds in the humanities , digital

humanities and computer science .

Project Topic Perfect Setup? R F H Major Project Tasks

Merseburg Incantations data retrieval, data modeling, geovisualization

Witchcraft Trials digitization, data modeling, geo-temporal visualization

Movie Relations network visualization, optimization

Actor Biographies interface design, network visualization, statistical charts

Musical Instruments geo-temporal visualization, uncertainty mapping

Letter Exchange text processing/ranking, parallel tag cloud variant

Engineering Subjects interactive tag cloud (to derive meta-subjects)

Musical Genres interactive tag cloud (to guide to close reading)

Instrument Makers repository linking, geo-temporal visualization

Witchcraft Trials II data modeling, geo-temporal visualization

Movie Locations interface design, geo-temporal visualization

Michelin Restaurants data retrieval, geovisualization, uncertainty mapping

Bundestag Politicians data retrieval, parallel coordinates variant

Music Styles data retrieval, geovisualization

Fake News data retrieval, text processing, tag cloud adaption

Hip Hop Musicians data retrieval, data modeling, network graph adaption

Education Spending data retrieval, geovisualization

Solidary Pact data retrieval, geovisualization

perience before they actually commit to join a digital

humanities project.

The theoretical part of the course covered half

of the semester. In order to practically apply the

gained knowledge, students required to implement

small project ideas in groups. I outlined the frame of

the practical training at semester start and asked the

students to think about projects they like to work on.

It was interesting to observe that the more time in the

theoretical section passed, the more precise students

were able to formulate project ideas using the learned

technical terms.

In three years, I faced very different student

groups. In the ﬁrst two years, only computer science

and a few humanities students were allowed to take

part in the course. In the last year, the ratio completely

switched as the course was offered for the journalism

and digital humanities Masters degrees for the ﬁrst

time. Thus, in all three years I had to improvise when

setting up interdisciplinary projects involving visual-

ization as well as humanities scholars. An overview

of all projects carried out in three years is provided in

Table 1.

4.1 The Perfect Setup

Only few projects involved students with a computer

science (or digital humanities) and a humanities (or

journalism or digital humanities) background alike.

This way, the whole project was in the students’

hands, and I just joined as a mediator if necessary.

The results of those projects were usually limited as

students faced coordination and communication is-

sues. In one case, humanities students very care-

fully generated a database manually, while, on the

other end, computer science students were waiting

for real world data to start with. Another issue arose

when only one student with a computer science back-

ground joined a group with more than one humanities

scholar—the computer science student was expected

to implement the project.

However, this constellation also brought forth re-

markable results. A project, in which one digital hu-

manities scholar with a decent computer science train-

ing and three journalism students participated, fo-

cused on the biographies of politicians of the German

Bundestag as of June 2019. They were inspired by the

New York Times visualization (NYT, 2019) reporting

on the career paths of politicians in the U.S. congress.

The students took the biographies from the Bundestag

homepage

having a far easier understandable visual-

ization as opposed to the New York Times in mind.

They focused on different aspects of a career: political

party, school degree, education degree, former politi-

cal position and former practised profession. In order

to extract those information for 709 politicians, they

implemented rules to parse the crawled biographical

texts accordingly. In order to browse the gained data,

they adapted a parallel coordinates visualization vari-

ant that can be used to interactively search for patterns

in the data. Figure 2 compares members of the SPD

and Die Gr

unen Bundestag factions. One can ﬁnd out

that all members of Die Gr

unen had a position on fed-

eral or state level before joining the Bundestag. In-

stead, many SPD politicians having a position in the

commune were elected for the Bundestag.

https://www.bundestag.de/abgeordnete/biograﬁen

Teaching on the Intersection of Visualization and Digital Humanities

103

Figure 2: A Perfect Setup project: Analyzing the biographies of 709 politicians of the German Bundestag.

Figure 3: A Perfect Setup project: Favored music styles in 10 German cities regarding the music features happy, dancable,

explicit, and energetic.

Another project group involving one computer

science and three journalism students focused on

the comparison of favored music features in East-

ern vs. Western German cities. Therefore, the Spo-

tify playlists

of ten cities were analyzed according

to the music features happy, dancable, explicit and

energetic. The mean value from a city’s playlist for

each of those features is mapped to the range [0, 1]

that is used to scale an emoji icon that reﬂects the cor-

responding music feature best. The resulting maps

are shown in Figure 3. As the students extracted the

playlists manually, they restricted their observation to

few German cities. They admitted that an appropri-

ate answer to the given comparative analysis interest

is hard to ﬁnd: the number of chosen cities was sub-

jective, the equal amount of representatives from East

and West did not stand for the population ratio, and

the number of songs in the playlists differed. How-

ever, the maps revealed some interesting information

on city level. Music heard in Frankfurt is typically

dancable and explicit, but less happy and energetic.

https://spotifymaps.github.io/musicalcities/

In Saxony, people listen to energetic music that is less

happy, dancable and explicit.

4.2 The Real Humanities Scholar

When too many computer science students partici-

pated in the course, I prepared a number of digi-

tal humanities projects that are usually driven by re-

search interests from humanities scholars. There-

fore, I asked my collaboration partners if they could

come up with a task that ﬁts into the practical train-

ing without overstraining the students. Moreover, I

asked them whether they could serve as collaboration

partners during the practical training to discuss pro-

totypes and results. The projects had a higher suc-

cess rate as opposed to The Perfect Setup as my col-

laboration partners are experienced in participating

digital humanities projects. Very good results were

achieved when a profound research query met an en-

gaged computer science student, also leading to pub-

lishing the outcomes (Meinecke and J

anicke, 2018;

Khulusi et al., 2020).

In two projects, students worked together with a

IVAPP 2020 - 11th International Conference on Information Visualization Theory and Applications

104

Figure 4: Real Humanities Scholar projects: Comparative visualization of geospatial-temporal metadata of instruments (left),

and tag cloud visualization supporting the interactive exploration of a composer’s music genres (right).

professor from the university’s musicology depart-

ment with whom I cooperate very successful since

2015 in various projects (J

anicke et al., 2016; J

anicke

and Focht, 2017; Khulusi et al., 2019). The ﬁrst

research interest focused on comparatively explor-

ing geospatial-temporal metadata of clavichord in-

struments in the catalogue of the Musical Instruments

Museum Leipzig. Next to mapping the clavichord

type to color, a visual design to communicate un-

certain datings and different geospatial granularities

was developed (see Figure 4, left). The second re-

search interest was an analysis of music genres of the

works of composers. Based on the musiXplora (Khu-

lusi et al., 2020) containing biographical information

about around 30,000 musicians, fulltext biographies

were scanned for a pre-deﬁned list of genre types

(such as opera, rondo, concert, etc.) and made avail-

able in an interactive tag cloud. Entering the name of

a composer, related genre tags are shown. The Infor-

mation Seeking Mantra is fulﬁlled as when clicking a

genre tag, the according text fragments can be close

read. An example for genre tags of Johann Sebastian

Bach is shown in Figure 4 (right).

4.3 The Fake Humanities Scholar

As the number of project ideas from my collabora-

tion partners was not exhausive enough, I—being a

computer scientist who worked in interdisciplinary

projects for ten years—acted myself as the humani-

ties scholar of some projects. Therefore, I prepared

project ideas that could generate interesting insights

into popular data sets. The project results in this

category were typically full-scale implementations of

the Information Seeking Mantra (Shneiderman, 1996)

serving undirected exploration purposes rather than

focused research interests.

The IMDb database

suited perfectly well in this

constellation as it provides various easily understand-

able information with different types, and only few

sophisticated visualizations (Vlachos and Svonava,

2013; Alam and Jianu, 2016; Auber et al., 2003) have

been proposed to explore ﬁlm history in the spirit of

the Filmﬁnder (Ahlberg and Shneiderman, 1994). Re-

lated student projects in my course included develop-

ing visual designs to explore movie relationships, bi-

ographies of persons such as actors or directors, and

ﬁlm locations. A screenshot of the tool developed to

browse ﬁlm locations, which had to be geocoded, is

shown in Figure 5. Each circle refers to one or mul-

tiple locations, and when choosing a location, related

movies can be explored. The bar chart encodes genre

with color, and a bar stands for the number of movies

ﬁlmed at the chosen location in a certain year.

4.4 The Helper

Especially in the last year with too few computer sci-

ence students in the course, humanities students re-

quired technical support. Many of those were journal-

ism students with only very basic computer science

skills. Partially, I trained them how to apply existing

tools and visualizations to their data. Also, students

took charge of services provided by the university for

computer science students of early semesters. Typical

for those projects was a very high quality of the data

source being composed, and that visualizations were

mostly adapted to ﬁt this data. In one project, the stu-

dents developed a database of German hip hop mu-

sicians, and they generated a list of relations among

https://www.imdb.com/interfaces/

Teaching on the Intersection of Visualization and Digital Humanities

105

Figure 5: A Fake Humanities Scholar project: Exploring

movie locations.

Figure 6: A Helper project: Social network visualization of

German hip hop musicians visualized with Gephi.

them. A close relation was deﬁned when two musi-

cians were participating the same band, a loose rela-

tion was deﬁned when two musicians having a similar

age were living in the same city. Gephi (Bastian et al.,

2009) was used for visualizing different subsets of the

graph as shown in Figure 6.

Another group focused on the history of German

Michelin restaurants. They were able to generate an

aestehtic visual output in the form of an interactive

map accompanied with a time slider that allows to

explore changes throughout the years (see Figure 7).

The star size encodes the number of Michelin stars

(1, 2, or 3) and a star is highlighted in orange once the

number of stars change. When transforming histori-

cal data that was provided in lists of Michelin restau-

rants per year, the students had to encode uncertain-

ties as some restaurants were already closed leading

to geocoding errors. As the main focus of the project

was comparing the coverage of Michelin restaurants

in Eastern and Western Germany, in such cases, cities

were chosen instead of exact positions keeping global

relationships stable. To support the comparison task,

Figure 7: A Helper project: Geospatial-temporal analysis

of German Michelin restaurants.

numbers of Michelin restaurants could be compared

by state using a horizontal bar chart.

In one case in the ﬁrst two years, I needed to re-

place the computer science students of a project as

they resigned from the course. The project focused

on how the Merseburg incantations are referred to in

the media. The collected data was very well struc-

tured and together with humanities students this lead

to a poster presentation at the digital humanities con-

ference in 2017 (J

anicke, 2017).

5 REFLECTION

In the three years I used different means to evalu-

ate the course, ranging from standardized to content-

related questionnaires. In addition, I received valu-

able feedback in discussions. Together with my ex-

periences gained, I reﬂect on important aspects of de-

signing a related course.

Group Heterogeneity. The focus of the theoreti-

cal course contents should be adjusted according to

the needs of the participating students. In the last

year, when the majority of students had a background

IVAPP 2020 - 11th International Conference on Information Visualization Theory and Applications

106

in journalism or digital humanities, I frequently dis-

cussed visualizations published in media. However,

some topics of the course were seen as too deeply

discussed (e.g., close reading, or temporal data visu-

alizations based on space-time cube operations (Bach

et al., 2015)). On the other hand, computer science

students, being the majority in the ﬁrst two years,

liked the comprehensive overview of related projects

from the digital humanities as an application domain.

Thus, on the one hand, working with a homogeneous

group makes it easier to adapt to the students’ needs,

on the other hand, a heterogeneous group ensures that

students with different backgrounds work together.

Student Projects. Students were seemingly more

engaged when they were able to choose their own

project topics. To longer the course took during the

semester, the more they were able to generate ideas on

their own—especially, journalism students brought

forth projects that were interesting from a technical

data analysis perspective while being relevant to so-

ciety too. I judged if their ideas ﬁt into the course,

and if so, we discussed the extent of the result to be

expected. Adaptions in this regard were made when I

recognized that a group performed better or worse as

expected when working on the projects. As projects

were carried out on the basis of Munzner’s nested

model (Munzner, 2009), which inheres iterative eval-

uation due to the involvement of domain experts in

the visualization design process, the evaluation of vi-

sualizations was not explicitly discussed.

Grading. Projects were not graded, but a successful

project was necessary to join an oral exam. Students

presented their project results in front of the group,

and they were asked to contextualize their project in

related works from visualization and digital human-

ities. I further engaged them to reﬂect on collabo-

ration experiences gained during their projects. In

three years, a total of four students resigned from the

course. Out of 18 projects, 13 were assessed as suc-

cessful, the students who presented insufﬁcient results

received additional project time. In this second itera-

tion 4 out of 5 projects were assessed as successful.

All students passed the subsequent oral exam.

Data Retrieval and Modeling. Many projects

faced data acquisition problems. On the one hand,

this relates to limited data processing skills, on the

other hand, to the short project durations. While in

one project numbers needed to be manually extracted

from old documents for which OCR failed, other

projects suffered from the long response time of data

deliveries—seemingly all valuable lessons learned for

future projects. Though students who start with al-

ready given data sets overall gained better project re-

sults concerning visualization quality, the challenges

of compiling and modeling a data set for a given re-

search inquiry is a valuable task, especially for (digi-

tal) humanities students. In addition, the exchange on

data modeling supports the interdisciplinary discourse

among students having different backgrounds.

Data Transformation. A broad overview of data

transformation techniques should be provided to con-

vey all aspects of the visual text analysis process suf-

ﬁciently. This was remarked by students as my lim-

ited lecture time allowed only the discussion of few

related methods such as Named Entity Recognition

(NER). Ideally, students should gain pre-requisites in

text processing before partaking the course.

Hands-on Sessions. Students without a techni-

cal background require practical hands-on sessions

throughout the theoretical part of the course. On

the one hand, they wish to apply existing visualiza-

tion tools to data sets—a common work practice of

digital humanities scholars and journalists—, on the

other hand, the programming of sample snippets was

seen valuable before beginning the practical training

in student projects. For the former, I suggest ses-

sions with standard tools like Voyant

, CollateX

Gephi

in order to foster understanding of the un-

derlying research questions and to discuss potential

visualization-related improvements.

Collaboration. Many students described the value

of the project work similar to the following remark:

“The development of visualization has brought me

a lot further, because I can now imagine the imple-

mentation of such projects much more concretely.”

Students with a non-technical background remarked

they would not have been able to design related

visualizations without the knowledge gained in the

course. Most of the students joined an interdisci-

plinary project for the ﬁrst time, and the course taught

them strategies how to approach them. Some students

valued the relevance of the project work for their fu-

ture, also they saw how the undertaken steps could be

applied to support other data analysis tasks. Lastly,

the exchange with other groups facing similar tasks

was important. On the contrary, students faced typi-

cal callenges of interdisciplinary work, especially in

the perfect setup projects.

https://voyant-tools.org/

https://collatex.net/

https://gephi.org/

Teaching on the Intersection of Visualization and Digital Humanities

107

An Ideal Course. A 5 ECTS course is too short

to teach all aspects of the visual text analysis pro-

cess (J

anicke et al., 2017), especially for students

having a non-technical background. I suggest two 5

ECTS courses in subsequent semesters with the fol-

lowing objectives:

• Semester 1—Contents: visual text analysis

overview, introduction to digital humanities, tra-

ditional vs. digitally supported workﬂows, text

mining basics, standard data transformation tech-

niques (e.g., tokenization, stemming, named en-

tity recognition, part-of-speech tagging)

• Semester 1—Project: data retrieval and data mo-

deling, preferably driven by an own project idea

• Semester 2—Contents: visualization design,

nested model, Information Seeking Mantra, close

and distant reading visualization techniques

• Semester 2—Project: iterative development of

a tool to support visual text analyses, preferably

based on a data set generated in the ﬁrst semester

Though the spread throughout two semesters in-

creases the likeliness of project groups breaking apart

or loosing members, this way, students gain more

time to generate data sets according to their own inter-

ests. If the ﬁrst project does not deliver an appropriate

data set, ready-to-use data sets can be used in the sec-

ond semester. As the course pursues that students ex-

perince interdisciplinary digital humanities projects,

only ”perfect setup” and ”real humanities scholar”

projects should be conducted. To avoid other con-

stellations, the ratios of partaking (digital) humani-

ties and computer science students should both range

around 50%, in case of a lower ratio of humani-

ties students, the number of real humanities scholar

projects should compensate this imbalance.

6 SUMMARY

The results illustrated in the ﬁgures are—without

any changes—the visual designs as presented by the

groups on the last course day. Even though most visu-

alizations leave room for improvements due to some

inappropriate design decisions, I was impressed by

the quality of project outcomes considering the fact

that many students had neither computer science nor

visualization skills before joining the course.

When designing a course like this, one needs to be

able to take different roles. For me, it was much easier

to take the helper role as my capabilities as a computer

scientist were requested. But my digital humanities

experience helped me also to join in as a fake human-

ities scholar. Nevertheless, having a series of project

ideas offered by real humanities scholars participating

as partners during the project developments is invalu-

able as it guarantees the steepest learning curve for

students with a computer science background.

REFERENCES

Ahlberg, C. and Shneiderman, B. (1994). Visual informa-

tion seeking using the ﬁlmﬁnder. In Conference com-

panion on Human factors in computing systems, pages

433–434. ACM.

Aigner, W., Miksch, S., Schumann, H., and Tominski, C.

(2011). Visualization of Time-Oriented Data. Springer

Publishing Company, Incorporated, 1st edition.

Alam, S. S. and Jianu, R. (2016). Analyzing eye-tracking

information in visualization and data space: from

where on the screen to what on the screen. IEEE

transactions on visualization and computer graphics,

23(5):1492–1505.

Andrews, T. L. and Mac

e, C. (2013). Beyond the tree of

texts: Building an empirical model of scribal variation

through graph analysis of texts and stemmata. Literary

and Linguistic Computing, 28(4):504–521.

Auber, D., Chiricota, Y., Jourdan, F., and Melanc¸on, G.

(2003). Multiscale visualization of small world net-

works. In IEEE Symposium on Information Visualiza-

tion 2003, pages 75–81.

Bach, B., Dragicevic, P., Archambault, D., Hurter, C., and

Carpendale, S. (2015). A Descriptive Framework

for Temporal Data Visualizations Based on General-

ized Space-Time Cubes. Computer Graphics Forum,

36(6):36–61.

Bastian, M., Heymann, S., Jacomy, M., et al. (2009). Gephi:

an open source software for exploring and manipulat-

ing networks. ICWSM, 8:361–362.

orner, K., Eide, O., Mchedlidze, T., Rehbein, M., and

Scheuermann, G. (2019). Network Visualization in

the Humanities (Dagstuhl Seminar 18482). Dagstuhl

Reports, 8(11):139–153.

Brehmer, M., Lee, B., Bach, B., Riche, N. H., and Mun-

zner, T. (2017). Timelines Revisited: A Design Space

and Considerations for Expressive Storytelling. IEEE

Transactions on Visualization and Computer Graph-

ics, 23(9):2151–2164.

Burke, B. (2012). A Close Look at Close Reading: Scaf-

folding Students with Complex Texts.

Busa, R. (1980). The Annals of Humanities Computing: The

Index Thomisticus. Raccolta delle opere di Roberto

Busa. North Holland Publishing Company.

Cao, N. and Cui, W. (2016). Introduction to Text Visualiza-

tion. Atlantis briefs in artiﬁcial intelligence. Atlantis

Press, Paris.

Cheema, M. F., J

anicke, S., and Scheuermann, G. (2016).

AnnotateVis: Combining Traditional Close Reading

with Visual Text Analysis. In Workshop on Visual-

ization for the Digital Humanities, IEEE VIS 2016,

Baltimore, Maryland, USA.

IVAPP 2020 - 11th International Conference on Information Visualization Theory and Applications

108

Collins, C., Viegas, F. B., and Wattenberg, M. (2009). Paral-

lel tag clouds to explore and analyze faceted text cor-

pora. In 2009 IEEE Symposium on Visual Analytics

Science and Technology, pages 91–98.

Dohn, N. and Dolin, J. (2015). Research-based teaching,

pages 43–63. Samfundslitteratur, 1. edition.

Dykes, J., Keefe, D. F., Kindlmann, G., Munzner, T., and

Joshi, A. (2010). Perspectives on Teaching Data Visu-

alization. IEEE VisWeek 2010.

Haentjens Dekker, R., van Hulle, D., Middell, G., Neyt, V.,

and van Zundert, J. (2014). Computer-supported col-

lation of modern manuscripts: Collatex and the beck-

ett digital manuscript project. Literary and Linguistic

Computing, 30(3):452–470.

Healey, M. (2005). Linking research and teaching explor-

ing disciplinary spaces and the role of inquiry-based

learning. Reshaping the University: New Relation-

ships between Research, Scholarship and Teaching,

pages 67–78.

Hinrichs, U., Alex, B., Clifford, J., Watson, A., Quigley,

A., Klein, E., and Coates, C. M. (2015). Trad-

ing Consequences: A Case Study of Combining Text

Mining and Visualization to Facilitate Document Ex-

ploration. Digital Scholarship in the Humanities,

30(suppl 1):i50–i75.

anicke, S. (2016). Valuable Research for Visualization and

Digital Humanities: A Balancing Act. In Workshop

on Visualization for the Digital Humanities, IEEE VIS

2016, Baltimore, Maryland, USA.

anicke, S. (2017). On the Impact of the Merseburg Incanta-

tions. In Conference Abstracts of the Digital Human-

ities 2017.

anicke, S. and Focht, J. (2017). Untangling the social net-

work of musicians. In Conference Abstracts of the

Digital Humanities 2017.

anicke, S., Focht, J., and Scheuermann, G. (2016). Interac-

tive visual proﬁling of musicians. IEEE transactions

on visualization and computer graphics, 22(1):200–

209.

anicke, S., Franzini, G., Cheema, M. F., and Scheuermann,

G. (2017). Visual Text Analysis in Digital Humanities.

Computer Graphics Forum, 36(6):226–250.

anicke, S., Geßner, A., Franzini, G., Terras, M., Mahony,

S., and Scheuermann, G. (2015). TRAViz: A Visual-

ization for Variant Graphs. Digital Scholarship in the

Humanities, 30(suppl 1):i83–i99.

anicke, S. and Wrisley, D. J. (2017). Interactive Visual

Alignment of Medieval Text Versions. In 2017 IEEE

Conference on Visual Analytics Science and Technol-

ogy (VAST), pages 127–138.

Jockers, M. L. (2013). Macroanalysis: Digital Methods and

Literary History. University of Illinois Press, Cham-

paign, IL, USA, 1st edition.

Kehoe, A. and Gee, M. (2013). eMargin: A Collaborative

Textual Annotation Tool. Ariadne, 71.

Kerren, A., Stasko, J. T., and Dykes, J. (2008). Teaching

Information Visualization. pages 65–91. Springer.

Khulusi, R., J

anicke, S., Kusnick, J., and Focht, J. (2019).

An Interactive Chart of Biography. Paciﬁc Visualiza-

tion Symposium (PaciﬁcVis), 2019 IEEE.

Khulusi, R., Kusnick, J., Focht, J., and J

anicke, S. (2020).

musiXplora:Visual Analysis of a Musicological En-

cyclopedia. In Proceedings of the 11th International

Conference on Information Visualization Theory and

Applications (IVAPP).

Meinecke, C. and J

anicke, S. (2018). Visual Analysis of

Engineer’s Biographies and Engineering Branches. In

LEVIA 2018: Leipzig Symposium on Visualization in

Applications.

Mesmer, H. A. and Rose-McCully, M. (2018). A closer look

at close reading: Three under-the-radar skills needed

to comprehend sentences. The Reading Teacher,

71(4):451–461.

Michel, J.-B., Shen, Y. K., Aiden, A. P., Veres, A., Gray,

M. K., , Pickett, J. P., Hoiberg, D., Clancy, D., Norvig,

P., Orwant, J., Pinker, S., Nowak, M. A., and Aiden,

E. L. (2011). Quantitative Analysis of Culture Using

Millions of Digitized Books. 331(6014):176–182.

Moretti, F. (2005). Graphs, Maps, Trees: Abstract Models

for a Literary History. Verso.

Munzner, T. (2009). A nested model for visualization de-

sign and validation. IEEE Transactions on Visualiza-

tion and Computer Graphics, 15(6):921–928.

Munzner, T. (2014). Visualization Analysis and Design.

CRC press.

NYT (2019). New York Times: How Every Member Got

to Congress. https://www.nytimes.com/interactive/

2019/01/26/opinion/sunday/paths-to-congress.html

(Retrieved 2019-07-08).

Propp, V. (2010). Morphology of the Folktale, volume 9.

University of Texas Press.

Shneiderman, B. (1996). The Eyes Have It: A Task by Data

Type Taxonomy for Information Visualizations. In Vi-

sual Languages, Proceedings, pages 336–343.

Sinclair, J. and Cardew-Hall, M. (2008). The folksonomy

tag cloud: when is it useful? Journal of Information

Science, 34(1):15–29.

Spence, R. (2007). Information Visualization: Design for

Interaction. Pearson/Prentice Hall.

Tamassia, R. (2007). Handbook of Graph Drawing and

Visualization (Discrete Mathematics and Its Applica-

tions). Chapman & Hall/CRC.

Viegas, F. and Wattenberg, M. (2008). TIMELINES: Tag

Clouds and the Case for Vernacular Visualization. in-

teractions, 15(4):49–52.

Vlachos, M. and Svonava, D. (2013). Recommendation and

visualization of similar movies using minimum span-

ning dendrograms. Information Visualization, 12:85–

101.

Ware, C. (2012). Information Visualization: Perception for

Design. Elsevier.

Teaching on the Intersection of Visualization and Digital Humanities

109