musiXplora: Visual Analysis of a Musicological Encyclopedia

Richard Khulusi

, Jakob Kusnick

, Josef Focht

and Stefan J

anicke

Image and Signal Processing Group, Leipzig University, Leipzig, Germany

Museum of Musical Instruments, Institute for Musicology, Leipzig University, Leipzig, Germany

Department of Mathematics and Computer Science, University of Southern Denmark, Odense, Denmark

Keywords:

Biography Visualization, Music Visualization, Digital Humanities.

Abstract:

Making large sets of digitized cultural heritage data accessible is a key task for digitization projects. While

the amount of data available through print media is vast in humanities, common issues arise as information

available for the digitization process is typically fragmented. One reason is the physical distribution of data

through print media that has to be collected and merged. Especially, merging causes issues due to differences

in terminology, hampering automatic processing. Hence, digitizing musicological data raises a broad range

of challenges. In this paper, we present the current state of the on-going musiXplora project, including a

multi-faceted database and a visual exploration system for persons, places, objects, terms, media, events, and

institutions of musicological interest. A particular focus of the project is using visualizations to overcome

traditional problems of handling both, vast amounts and anomalies of information induced by the historicity

of data. We present several use cases that highlight the capabilities of the system to support musicologists in

their daily workﬂows.

1 INTRODUCTION

Traditional musicology has developed into a ﬁeld

with several sub-domains. They differ by their main

object of interest and also by their view on the data.

Examples are instrument making, dealing with the

physical production and often also restoration of in-

struments, or organology (Tresch and Dolan, 2013),

concentrating on methods of research, teaching, and

documentation of instruments. Further sub-domains

are prosopographical analysis, focusing on the per-

sons associated with musicology (biographical re-

search) rather than instruments. Additionally, some

musicologists focus on inspecting developments of

and inﬂuences on places important for music his-

tory. Thus, in musicology, many different types of

entities are encountered. Typically, these are classi-

ﬁed as persons, (musical) objects, institutions, places,

terms, events, media, and titles. This range of enti-

ties, combined with different sub-domains, leads to a

vast amount of musicology data to be handled. Espe-

cially, use cases and research questions with restricted

focuses subdivide the ﬁeld into well-researched and

less-researched fragments of data. Tools trying to

connect those data fragments are few, because tra-

ditional musicological approaches either do not need

to get a comprehensive picture of the whole musico-

logical knowledge—knowing in which location an in-

strument was produced being of less interest for a re-

storer compared to the three-dimensional instrument

model—and handling the vast data is hardly possible

by traditional means.

In this paper, we give give an overview of an interdis-

ciplinary collaboration between visualization scholars

and musicologists, aiming to link the different frag-

ments of musicological knowledge and offering them

to the broad public with the help of an online ex-

ploration tool—the musiXplora—, supported by dif-

ferent visualizations to allow usage for both experts

with speciﬁc research questions at hand as well as for

casual users interested in browsing musicological in-

formation. A screenshot of the system is shown in

Figure 1. As part of the digital humanities, we also

want to highlight development of a digital tool in the

ﬁeld of musicology, leading to both, new needs and

interests in research of The Musicology as a single

and comprehensive ﬁeld, and also possibilities aris-

ing through deploying computer science technology

and digitization, especially for accessing and linking

vast amounts of data.

2 RELATED WORK

Throughout the last two decades, a multitude of visua-

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

musiXplora: Visual Analysis of a Musicological Encyclopedia.

DOI: 10.5220/0008977100760087

In Proceedings of the 15th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theor y and Applications (VISIGRAPP 2020) - Volume 3: IVAPP, pages 76-87

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

Figure 1: The musiXplora showing different visualizations of instrument makers in the German city of Markneukirchen.

lization projects for musicology have been con-

ducted (Khulusi et al., 2020). Through these, dif-

ferent sub-domains and entities of musicology have

been analyzed with the aid of visual tools. Some

projects focus on a prosopographical inspection of

persons. While generic visualization tools for per-

sons (Leskinen et al., 2017) are used to show

domain-independent features such as networks (Lu

and Akred, 2018; Miller et al., 2012; Gleich et al.,

2005; Vavrille, 2017; Crauwels and Crauwels, 2018),

places (Doi, 2017) or temporal data (Andr

e et al.,

2007), musicological research often requires domain-

speciﬁc, contextual information. For example, re-

garding musical instruments (objects), typical visu-

alizations use three-dimensional data, gained through

CT (Borman and Stoel, 2009; den Bulcke et al., 2017;

Hopfner, 2018) or other techniques (Heller, 2017;

Konopka et al., 2017) to generate volume or surface

renderings (Tuniz et al., 2012). But the sub-domain of

organology does not only focus on the physical prop-

erties of instruments, other research questions are di-

rected toward contextual metadata like dates, places,

and ownerships of instruments, usually not processed

by these tools. Not only do single sub-domains suffer

from a too strict focus on a small set of information

for a speciﬁc kind of entity, but also from compre-

hensive and domain-wide inspection of data not being

provided. First ground building work for enabling a

more global resource of musicological knowledge has

been done by the German research project BMLO.

2.1 BMLO

Short for Bavarian Musician Lexicon Online (Bay-

erisches Musiker Lexikon Online) (Focht, 2006), the

project consists of a database and a website. A

group of musicological redactors crawled different

analog and digital resources (manually) and put them

into a standardized and scalable relational data base,

suitable for an automatic processing. Due to the

long history of musicology—and missing standards—

a (semi-)automated processing of data is mostly only

possible after manual preprocessing. The data col-

lected in this process includes the seven categories

deﬁned by the musicologists of the project: persons,

objects, places, institutions, terms, media, and events.

In the active years of the BMLO project (2004–2014)

the number of persons in the data set rose to around

30,000, while other categories (institutions, terms,

media, ...) were mostly only drafted. Nevertheless,

a frontend offers a search interface and list-like pre-

sentation of the collected data, enriched with pic-

tures crawled through Wikimedia. Next to typical bi-

ographical data, entities also list their identiﬁers on

different domain-related websites, like VIAF

, GND

or Q

(Wikipedia’s/Wikimedia’s identiﬁer), building

a linkage between the BMLO and other online knowl-

edge sources. In 2014, the BMLO project moved

from Munich to Leipzig University and was extended

to the musiXplora project, abolishing the focus on

Bavaria and collecting generally available data of mu-

sicology. A further focus of the project was the inclu-

sion of various interactive visual interfaces aiming to

support a broad range of musicological research tasks.

www.viaf.org

www.dnb.de/DE/Professionell/Standardisierung/

GND/gnd node.html

www.wikidata.org

musiXplora: Visual Analysis of a Musicological Encyclopedia

3 DATA

Like the BMLO, the musiXplora database (Focht,

2019) consists of the seven facets of musicology,

divided into different repositories and is editori-

ally maintained in the Musikinstrumentenmuseum der

Universit

at Leipzig

. Each repository appears in its

own color, used in the header, on Single Result Pages

as well as in the visualizations. Through the usage of

the BMLO data, challenges from the former project

were also inherited. These include the work with his-

torical, uncertain data, a vastness with potential for

distant-reading, but challenges for searching, ﬁlter-

ing and ﬁnding of speciﬁc entities, and the missing

of standards for general terms of musicology.

3.1 Musici—Persons

Musicologists may be interested in persons, associ-

ated in creating music—like musicians, composer or

singers—, but also in people with professions like

music instrument makers or restorers. The people

associated with music are collected in the Musici

(person) repository. Persons m

, m

, ..., m

, with n >

32, 000 offer a multitude of metadata for each musi-

cian m

1. m

Name

—Different written variants for ﬁrst and

last name, as well as additional names like

pseudonyms or maiden name

2. m

Date

—Temporal Data - Life and work years in a

range of [0,2019] A.D.

3. m

Con f ession

—List of Confession(s)

4. m

Gender

—Gender(s)

5. m

MusicalPro f essions

—List of musical professions

6. m

OtherPro f essions

—List of non-musical professions

7. m

Branch

—List of types of the branch(es) that em-

ployed the person like state, court or military

8. m

—List of musical institutions c

, ..., c

the per-

son belonged to

9. m

—List of places l

, ..., l

divided by type [Birth,

Death, Work, ...]

10. m

IDs

—List of IDs used in related repositories of

musicological knowledge [VIAF, GND, Q, ...]

11. m

Links

—List of links using the IDs to reference

other resources containing important data of the

person

https://mfm.uni-leipzig.de/

3.2 Casae—Institutions

Unions of persons are captured in the Casae reposi-

tory (Institutions), such as opera houses, court orches-

tras or festivals. For each institution c

, ..., c

informa-

tion, including the following, are given:

1. c

—List of members m

, ..., m

2. c

Name

—Name of the institution

3. c

Date

—Founding and closing date

4. c

—List of locations l

, ..., l

While, compared to the persons, the range of meta-

data is rather low, the connectivity from institutions

to persons and locations leads to great possibilities

for distant reading analyses of the institutions itself or

more general trends between multiple ones (Khulusi

et al., 2019).

3.3 Loci—Places

While not directly related to music, places are a fur-

ther aspect of interest for musicologists. As seen

above, places are for example included in data sets

of persons and institutions. Especially for centuries

prior to globalization and fast traveling possibilities,

different places have been centers of agglomerations

(see Section 5.2). The information for places l

, ..., l

with n > 76, 000, includes production centers for spe-

ciﬁc instruments or centers of performance, for exam-

ple at courts. Like the repositories above, given infor-

mation includes l

Name

, l

IDs

, l

Links

, and l

. Further,

topological information is given hierarchically with

Hierarchie

, linking e.g. l

Germany to l

Europe as

parent and to the German states l

, ..., l

as children.

More information is given with e.g.: l

Coordinates

—in

the form of longitude and latitude coordinates—or

with linkage to events l

3.4 Baccae—Objects

A further important part of music and musicological

research are the objects needed for or produced by

music (Baccae), including instruments and composi-

tions. b

, ..., b

have properties about the label b

Name

a categorization leading to a Res-entry r

, describing

textual data b

Description

and events b

Events

linking an

object to places l, persons m, temporal data b

Date

and

type of event b

Type

3.5 Res—Terms

The repository Res includes terms of musicology.

While normally dictionaries exist that help in under-

standing terms and offering descriptions, musicology

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

has the unique issue that a lot of terms are uncer-

tain, thus, leading to challenges for automated pro-

cessing (Khulusi et al., 2020; Kusnick et al., 2020).

This follows from changes in the meaning of terms

throughout the time and almost no endeavors to stan-

dardize terminology. While in certain centuries a spe-

ciﬁc string instrument may have had ﬁve strings, it

may have been overhauled in later years and adopted a

six-string construction, still keeping its original name.

This leads to the need for more context information

to analyze written musicological information and (es-

pecially for automatic processing) issues in merging

different data sources. Due to these issues, that are

still aspects of the present research, the Res repos-

itory tries to collect different descriptions under the

same name. For this purpose, r, as the set of r

, ..., r

contains r

Variants

—labels in different languages or

synonyms—, describing elements r

Description

, lists of

objects of this kind r

and links to other resources

holding information about this kind of term r

Links

3.6 Catalogus—Media and Titles

While Baccae contains physical objects, the digital

representatives of objects (books’ metadata, objects’

3D data or contents of CDs or Books) are collected in

Catalogus. While this repository is to be named in the

collection of repositories, it does not have a sufﬁcient

state of research to be further discussed in this paper.

3.7 Eventa—Events

As seen before, Eventa entries e

, ..., e

offer connec-

tions between the different facets. Each event e

con-

tains information about its type e

Type

, temporal com-

ponent e

Date

, associated person e

or object e

and

place e

4 FEATURES

With the goal to offer a digital knowledge base for

musicology, the musiXplora has different core fea-

tures, taking inspiration from traditional sources and

enriching them with computer science technology.

These goals—or tasks—ﬁt into Munzner’s Task Clas-

siﬁcation Scheme (Munzner, 2009).

4.1 Searching

The most obvious feature is offering a faceted search,

enabling a multitude of ways to ﬁnd entities. Mun-

zner’s second task category Searching categorizes

Figure 2: Four different example searches for persons. Dif-

ferent kinds of data (token, ID or date based) and different

logic operators can be used.

such tasks by the awareness of target and its location.

The musiXplora supports all four subtasks. On the

one hand, a known entity can be found with the use

of the speciﬁc entity’s x

features x

(like x

Name

), called Lookup or Locate. On the other, a search

rather detached from a speciﬁc entity can be started

by searching for feature-value pairs with high recall

like the century of birth or profession (Exploring or

Browsing). Locate and Explore are deﬁned with an

unknown location of the target. For our search, this

is reﬂected by a search not resulting in a single en-

tity x

but a set of results x

, ..., x

, with n > 1. For

such scenarios, the next section will provide further

information (see Section 4.2).

Additionally, for the search-ability, wildcards and

logic operators are included, helping in deﬁning rel-

evant parts of values (e.g. ranges of dates or groups

of entities). A so-called Simple Search is offered as

default. This matches the inserted search tags to a

predeﬁned reverse index containing all IDs to each

possible tag of an entity. This reverse index allows

a quick search and high recall, while the precision

is rather low, as tags may be ambiguous (a tag may

be the name of a person or a location). For cases

where the user needs a high precision or knows ex-

actly what to look for, a combination of type and to-

ken can be searched for. In simple cases, this may be

First Name and Wolfgang. Different inputs may also

be combined. A search for First Name: Wolfgang%

and Place of Work: Vienna gives 20 results (compared

to 243 for just Wolfgang). Also, logical operators like

AND (space), OR (Pipe) and NOT (Tilde) are possi-

ble and wildcard operators (percentage mark for mul-

tiple chars and underscore for a single char) are of-

fered. Figure 2 shows four different example searches

by users of the musiXplora. Different use cases exist,

where this kind of search is not sufﬁcient to ﬁnd the

entity needed. May it be caused by only uncertain

knowledge of the entity (the name only being phonet-

ically known, but not in the written form), only vague

knowledge or an indirect search interest. This leads

musiXplora: Visual Analysis of a Musicological Encyclopedia

Figure 3: A timeline of working times (darker blue), en-

riched with living time (lighter blue). Different shapes indi-

cate temporal uncertainties in the data.

to searches having high recall and low precision (e.g.

a search only by the place of work, often resulting

in multiple thousands of results). In analog dictio-

naries, this problem also exists, but on a lower scale,

as such directories mostly return a much smaller data

set and do not offer interactivity to follow up on such

a search. This is caused by limited space and also

by most directories being specialized on a speciﬁc

set of entities (e.g. musicians or instrument makers).

Hence, with a wider set of data this problem is more

crucial. To tackle this issue already encountered in the

BMLO, distant- and meso-reading visualizations are

included in the musiXplora, helping in distinguishing

the entities returned in a search.

4.2 Visualizations, Overview and Filter

These visualizations depict the distribution and range

of features for all found entities. The inclusion of

those is a real improvement of the musiXplora in con-

trast to the rather dictionary-like implementation of

the BMLO and hold great value for the users of the

system. In general, all visualizations follow Shnei-

derman’s Mantra (Shneiderman, 1996) and offer a

full overview as default and means for ﬁltering the

data set on demand. For the above-mentioned search

of First Name: Wolfgang% and Place of Work: Vi-

enna the following visualizations will be offered. For

temporal data, a timeline is used, dynamically adjust-

ing to the range of dates and stacking entities above

each other. These timelines communicate the uncer-

tainty of dates with shapes (arrow-like for time ranges

and straight edges for time stamps) and a differen-

tiation between lifetime (low saturation) and docu-

mented years of work (high saturation) as seen in Fig-

Figure 4: Glyph based map, showing different kinds of

places with different colors. Heavily overlapping circles are

aggregated to pie charts. The digit indicates the number of

aggregated entities.

Figure 5: Different examples of pie charts and sunbursts to

grasp the distribution of features in the search result.

ure 3. Geo-spatial data is shown in a glyph based map,

where the different types of places—like the place

of birth/death/work/... for persons and the construc-

tion/restoration/changing ownership/... for objects—

are coded with different colors. The problem of over-

lapping by giving a full overview is met with ag-

gregating glyphs. To combine glyph and color ap-

proaches, each glyph aggregating more than one en-

tity is depicted as a pie chart, showing the distribution

within. Figure 4 presents dozens of places of work in

Europe, with a couple in America and a single occur-

rence in the Middle-East for our example search.

We also use pie charts’ ability to convey distribu-

tions to show more general feature distributions like

seen in Figure 5. For this, multiple pie charts are

added as standalone visualization at the bottom of the

page. For persons m

, ..., m

, these may be different

MusicalPro f essions

, m

Non−MusicalPro f essions

, m

Institutions

Con f ession

, m

Gender

and others. Similar, for the other

repositories, different pie charts show distributions of

their speciﬁc features. For some of these features, hi-

erarchical information is given. All places have topo-

logical information upwards and downwards (earth -

continent - country - state - district - city - quarter)

placing them in a hierarchical context. Also, profes-

sions (e.g. Singer - Soprano/Alto/Tenor/...), Confes-

sions, and more are given hierarchically. As pie charts

do not allow for hierarchical information, we adopted

the sunburst plot for these features. Figure 5 shows

example pie charts for the search above and the mu-

sical professions as sunburst in the lower-left corner.

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

Figure 6: A force-directed graph with each person of the

result being initially placed on the center of the y-axis and

linearly distributed on the x-axis. Lighter nodes indicate

persons added as context and not being part of the actual

result.

Technically, the segments on the inner ring are dis-

tinct categories (singer, instrument maker, ...) and, if

available, child elements will span on an outer ring,

bearing the name (soprano, alto, tenor, ...) and the

amount of those sub-types. Although deeper hier-

archies are available, we only allow for up to three

subsections, as each subdivision reduces the available

space for labels drastically, which can also be seen in

Figure 5.

In the case of relational data, a network graph

with a force-directed layout is included (see Figure

6). Typical for this kind of visualization, dots rep-

resent entities and relations among them are shown

by connections. Color is used to communicate dif-

ferent types of relations. In contrast to the other vi-

sualizations, we add contextual data in the visualiza-

tion that is not included in the actual search result.

This helps in getting insight into an entity’s connec-

tivity. Such contextual information has the form of

additional nodes mc

, ..., mc

showing the ﬁrst level

neighbors of the observed entities. Additional enti-

ties have a reduced saturation and clicking is deacti-

vated, which all other visualization elements provide

to select a single or a range of results for ﬁltering. As

observable in the different ﬁgures above, some issues

with the visualizations still exist. Due to the usage of

multiple visualizations at once to give an exhaustive

overview of the search results, less space is available

for each single visualization. While this lowers the

level of detail (e.g. as visible in Figure 5 in which too

small labels are hidden), we tackle this problem by

enabling interactions like tool tips containing full in-

Figure 7: Single Result Page of Wolfgang Amadeus Mozart,

listing different name variants, dates and further biographi-

cal information (left) and visualizations for these (right).

Figure 8: Start Page Visualization of Musici showing the

persons with dates of birth or death at the current day.

formation (all labels for a hovered pie chart slice, all

places aggregated under a map glyph, ...). For tasks

requiring a more detailed inspection of results, we of-

fer alternative visualizations.

4.3 Finding

After the searching part, the user may click on a single

entity x

or ﬁlter until only a single one is left. This

can either be done in the result list or through one of

the visualizations. The right side of the page is now

changed to a Single Result Page (Figure 7). Here, all

available information listed in the data section (Sec-

tion 3) are shown in textual form.

Differently from the previous pages, we now de-

ploy more close-reading like visualizations, concen-

trating on the selected entity x

and offering a com-

prehensive overview. These visualizations are placed

next to the textual information as small visualization

previews. On demand, a user can click on each vi-

sualization to get a full view (seen later in the use

case Section 5). This presentation allows for more

details due to focusing on a single visualization at a

time, in contrast to multiple ones in the searching sec-

tion. While geo-spatial data of persons, places, and

objects are again shown on a glyph based map and

relations using a graph, novel visualizations are for

example deployed for institutions. To allow explo-

ration of temporal developments of an institution An

Interactive Chart of Biography (Khulusi et al., 2019;

Khulusi et al., 2018) is included for each institution.

This visualization links membership data of the in-

stitution to a time-axis and offers distant-, meso- and

close-reading views on the data, allowing it to be used

for a broad range of scenarios and research questions

regarding temporal developments of institutions. All

included close-reading visualizations differ not only

through the focus on a single entity but also in terms

of interaction. While their distant- and meso-reading

counterparts allow ﬁltering and selecting for ﬁnding

entities of research interest, the Single Result Page

musiXplora: Visual Analysis of a Musicological Encyclopedia

focuses on linking entities database-wide. As an ex-

ample, all person’s locations m

link directly to their

representation in the Loci places directory, indicated

by a coloring of the label in the repository’s color

on mouse-over. Hence, quick navigation between the

repositories and entity category is enabled on a high

level of detail, which will be discussed in-depth in

Section 5.

4.4 Browsing

Besides Searching and Finding, an explicit Browsing

of the data is also offered. While this is less useful for

speciﬁc use cases and research inquiries, it offers in-

teresting and easy access to the data, especially useful

for casual users. To support this task, the start page of

each repository offers a Start Page Visualization, be-

fore a search can be performed. The visualized data

is selected according to the opened repository. For

persons, those are shown having the date of birth or

death at the current day (see Figure 8). For places, of-

ten searched entities are shown, or a random selection

for all other repositories. The different visualizations

include a force-directed graph layout for persons, ob-

jects, institutions and media, a tag cloud for terms, a

map for places and a timeline for events. If available,

entities are grouped, e.g. persons being grouped by

the type of anniversary, institutions by their location

(on a city level) and objects by their instrument type.

Figure 9: Single Result Page for Ludwig van Beethoven, as

only result for the string ”Beethoven” for a Simple Search.

5 USE CASES

During our collaboration with musicologists, we ob-

served their use and utilization of the musiXplora. For

this, multiple evaluation meetings where held, where

needs and general feedback of the main collaborating

musicologist were collected and discussed. Later, we

observed daily workﬂows of musicologists of the Mu-

sic Instruments Museum of Leipzig University and

how the musiXplora was used as a research tool. As

Figure 10: The network of Beethoven as visualization

(right), enriched with a categorized listing (left). Instrument

maker Nanette Streicher-Stein is selected and highlighted.

expected, the tool itself did not show a speciﬁc use

case of how it is used, but rather a multitude of dif-

ferent uses for the (different) experts. To elaborate on

this, we discuss three different use case types.

5.1 Information Lookup

The most basic use case observed by musicologists

working with the tool was a simple information

lookup, leading to a more specialized question. In

the discussed case, the musicologist wanted informa-

tion about Ludwig van Beethoven’s instrument mak-

ers and associated instruments. For this, the expert

started by querying the search interface. Through this

search, we only get a single result, as Ludwig van

Beethoven is the only person with the tag Beethoven

in the database. As a well-known person in musicol-

ogy, the resulting site shows a well-researched state

and gives access to a multitude of information as seen

in Figure 9.

In the next step, the researcher observed the social

network of Beethoven. Figure 10 shows the entire

network, accessed through clicking on the network

Figure 11: Single Result Page for the instrument maker

Nanette Streicher-Stein.

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

Figure 12: Baccae Single Result Page for a ”Ham-

merklavier”, produced by Streicher-Stein and located in the

Musikinstrumentenmuseum der Univerist

at Leipzig (Music

Instrument Museum, Leipzig University).

graph preview on the right side of Figure 9. In this

close reading visualization, different colored edges

differentiate the various kinds of relations. Through

this, the user ﬁnds famous instrument makers like the

married couple Johann Andreas Streicher and Nanette

Streicher-Stein (highlighted) as part of Beethoven’s

network. A click on the latter leads us to the Sin-

gle Result Page of Streicher-Stein seen in Figure 11.

Next to the information about her relations, we also

have access to information like different name vari-

ants known for her, in this case, her other ﬁrst names

Maria and Anna. Further, her professions list piano

player, piano and organ maker, and more. We also

get her informal and contemporaneous titles, like ”Di-

rector of the Piano-Factory”. Below the green bio-

graphical entries, we ﬁnd the list of links to exter-

nal sources in red, including Wikipedia, Wikimedia,

BMLO, MusikerProﬁling (J

anicke et al., 2016), AKS

and further important sources for musicology. With

the help of the also listed Musical Instrument Muse-

ums Online (MIMO), we can query the centralized

database of instrument museums and their informa-

tion about Streicher-Stein with a single click. This

leads us to a list of instruments produced by her.

One of these instruments can be seen in the screen-

shot of the Baccae repository of the musiXplora in

Figure 12, presenting a brief overview of the instru-

ment with the hint that the object’s page is still un-

der construction (which is the cause why the instru-

ment is not yet directly linked on Streicher-Stein’s

page). Further, we can access a list of all instruments

by Streicher-Stein exhibited in European museums.

This list consists of only a dozen pianos and con-

tains only known and documented instruments made

by Streicher-Stein. While the actual number of pro-

duced instruments was without a doubt higher, the

number of instruments produced per city was signiﬁ-

cantly lower than in the following centuries.

5.2 Interest Browsing

The prior use case on a famous instrument maker

prior to the 19th century arouse the musicologist’s in-

terest in the development of the profession. He looked

at the general state and development of instrument

makers appearances in the database and ﬁltered for

only those that had a place of work in Vienna, which

includes Streicher-Stein (see Figure 13, top). For

higher precision, the expert then included only those

persons with the main place of activity in Vienna (see

Figure 13, bottom). An increase of the number of

instrument makers around the change of century can

be seen. Having musicological expertise, the expert

concluded that the beginning of the Industrial Revo-

lution (in Germany around 1815-1835) brought fun-

damental changes to the production, availability and

the demand of music instruments in Europe, causing

the beginning of a reshaping of the profession from

straight craftsmanship to mass production. Further,

the expert reﬂected hypotheses on abstract concepts

inﬂuencing musicology, although not directly linked

to it but rather to social or cultural aspects. Simul-

taneously to their existence, tools and means of vali-

dation and showing such concepts are missing in tra-

ditional musicology and a clear need for such tools

exists. In recent years, ﬁrst visualization approaches

dealt with such abstract concepts like chauvinism or

historism (Khulusi et al., 2019) and how they are re-

ﬂected in musicology.

After the browsing of instrument makers in Vi-

enna in the 18th century, the musicologist directed his

investigation towards the inﬂuence of the Industrial

Revolution on instrument making in Central Europe.

The hypothesis is an increased demand for mu-

sic instruments due to easier transportation means and

lower prices of instruments, caused by a combination

of easier accessibility of raw material and division

of labor (not all parts of the instruments were con-

structed by the same instrument maker). Figure 13

supports this assumption for Vienna, as the number

of instrument makers increased at ﬁrst, which can be

seen as an indication of an increase in demand. Fur-

ther, a view on the instrument makers in Central Eu-

rope in general (Figure 19) shows that the rising trend

continues throughout this time. Hence, the visualiza-

tion of the data supports the hypothesis without refer-

ring to instruments, for which currently data is rarely

Figure 13: Timeline showing all instrument makers with

a general place of activity (top) or main place of activity

(bottom) in Vienna.

musiXplora: Visual Analysis of a Musicological Encyclopedia

Figure 14: Result for the search of all persons with the main

place of activity in F

ussen or Mittenwald.

at disposal. Thus, using the occurrence of instrument

makers to deduct trends for the instruments turned out

to be a working strategy of the expert.

Next to such economic developments, the Indus-

trial Revolution brought a wide set of technological

advances shaping the every-day life. An example is

given with the invention and expansion of the train

and the train network. This led to a change in the local

production center’s location characteristics. Centers

of high inﬂuence and renown like Vienna lost their

standing if they did not ”jump on board” of those ad-

vances.

Especially traditional centers like F

ussen and Mit-

tenwald in Bavaria suffered from a late connection

to the German and Central European train network.

In Figure 14, all persons whose main working place

was in at last one of these two cities are shown. The

temporal development shows that the production cen-

ters were in a growth phase that abruptly stopped

with the beginning of the Industrial Revolution in

Germany (ca. 1815-1835). Close to the German-

Austrian border, both locations were close to impor-

tant courts and cities at the time and their proxim-

ity to rivers (Isar and Lech, respectively, both join-

ing, the Donau) allowed early transportation means.

The 19th century decrease of instrument makers in

ussen and Mittenwald, that is visualized, is accom-

panied with an increase of instrument makers in other

cities, e.g., Markneukirchen at the German-Czech

border. Markneukirchen suddenly developed to one

of the most important centers in Germany, produc-

ing all kinds of instruments and even shipping them

Figure 15: Top of the Single Result Page of Loci for

Markneukirchen.

Figure 16: Overview of all instrument makers with main

place of activity in Markneukirchen.

intercontinental. Figure 15 shows the Loci page for

Markneukirchen, giving us direct access to all persons

with either main place of activity (Figure 16) or gen-

eral place of activity in the city (Figure 17).

In Figure 18, the temporal development of the

three locations is juxtaposed. Instrument makers of

ussen and Mittenwald are shown at the top, and the

ones having Markneukirchen as main place of activity

are shown at the bottom. Additionally, this transition

of instrument makers from former important centers

to new ones is supported by a view of all instrument

makers in Europe. It shows only a shallow differ-

ence in the total number of instrument makers (see

Figure 19). A nearly constant amount, paired with a

decrease in traditional centers, indicates a transition

of locations. In the second half of the 20th century,

a sudden drop in the number of instrument makers

is seen for Markneukirchen and Europe in general.

One reason is that fewer people are included in the

database in the past decades as people are mostly too

young to be having a meaningful impact on musicol-

ogy. Nevertheless, actual events also reinforced this

development. Economically, the advance of division

of labor and political nationalization in the German

Democratic Republic lead to a vanishing of instru-

ment makers’ names as a kind of brand.

Figure 17: Overview of all instrument makers with place of

activity in Markneukirchen.

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

Figure 18: Match up of all persons with the main place of

activity in F

ussen or Mittenwald (top) or Markneukirchen

(bottom) over a synchronized time-axis.

5.3 Exploration

Casual users typically access a repository through its

Start Page Visualization (see Section 4.4). We take a

look at popular places in Loci, the location dictionary.

Figure 20 shows a few randomly selected places, from

users’ frequently accessed places. In the image, we

can see places in the USA, Russia, Ukraine, Czech

Republic, Germany, Italy, France, and Spain. Inter-

ested in the places of southern Germany, Swiss, Aus-

tria and Italy, we zoom in to get a more precise view

on the glyphs (Figure 21). From this selection, we ac-

cess the Single Result Page of Milan in northern Italy

(Figure 22). Next to the precise topological informa-

tion of Milan, located in Lombardy, we can see all

links to persons through their different kinds of activ-

ity in the city and a single institution being located

here. Next to the listing of entries associated with

the links, we can also switch to the Musici (person)

repository by selecting an activity of interest. To get

an impression on the people strongly belonging to Mi-

lan, we select the Main Place of Activity label with its

24 results.

Musici’s overview page of these 24 persons con-

sists of the visualizations seen in Figure 23. While

24 persons are too few to deduct general trends and

developments of groups of persons like in the sec-

ond use case, this overview may still hold informa-

tion of interest for a browsing task and show move-

Figure 19: Timeline of all instrument makers in Europe and

their temporal occurrence. Nearly constant state during the

years 1800 to 1930 supports hypothesis of transitioning of

production centers seen in Figure 18.

ments of people that used to work mainly in Milan.

The network graph of this search (see in the middle of

Figure 23) shows low connectivity for these persons.

Only the ﬁrst component consists of more than one

person of the search result (two black nodes, while

gray ones indicate persons, not included in the result).

Indicated by the purple edge, both persons are of the

same kin. A click on the nodes shows us the informa-

tion that these two persons are rather unknown (only

a sparse amount of data is given) and that they had a

familial relationship (uncle and nephew). The last set

of visualizations consists of pie charts and sunbursts

(Figure 23, bottom). At ﬁrst glance, we notice a quite

large percentage of employers (branch) of the instru-

ment making section. The sunburst of musical profes-

sions adds information about the actual kind of pro-

fessions by these persons. If we would be interested

in getting a full overview of all these instrument mak-

ers, we could click on the Instrument Making section

of the Branch sunburst or on one of the segments of

the Musical Professions sunburst to set a ﬁlter to only

those persons included in the categories. Other pos-

sible paths leading to further inspections could be for

example to view the persons in one of the three listed

institutions or persons having speciﬁc non-music re-

lated professions like statesman, painter or librarian.

Figure 20: Start Page Visualization for Loci, showing a ran-

dom selection out of the most user searched locations.

6 CONCLUSION

With the growing importance of digital methods in the

humanities, the amount of data digitally available is

likewise rising continuously (Windhager et al., 2018;

anicke et al., 2015). Due to the nature of the ﬁeld,

available data in musicology is typically fragmented

due to different, loosely connected sub-domains. A

single fragment bears a vastness of data but mostly fo-

cuses on a speciﬁc view on the ﬁeld. Global trends in

musicology can only be hypothesized, and only a few

tools and projects give a comprehensive overview.

The musiXplora is an on-going project dealing with

collecting, standardizing and visualizing information

of seven different types of data deﬁned by musicol-

musiXplora: Visual Analysis of a Musicological Encyclopedia

Figure 21: Zoomed in view on the Loci start page.

ogists. Due to close collaboration between visual-

ization scholars and musicologists, the tool does not

only make use of digital advantages of data storage

and the capability of visualization to make data eas-

ily accessible, but also guarantees correctness, qual-

ity, and relevance of the content for both, experts and

casual users. Different use cases showcase the abil-

ity of the system to be used for a wide set of research

questions and also enable different ways of access-

ing resources. Thus, the experts could validate a hy-

pothesis of the abstract concept of the Industrial Rev-

olution and its inﬂuence on music instrument makers

through visualizations of places and persons. Further,

we observed a large interest and fascination during

the usage of the tool and through on-going interdisci-

plinary discourses on the needs and interests of differ-

ent musicologists. While the tool is generally tailored

for the musicological data set, other projects using

this data (J

anicke et al., 2016; Meinecke and J

anicke,

2018) have already shown that an adaption to non-

musicological data is possible, which offers diverse

opportunities for future research.

Figure 22: Single Result Page of Milan, including all per-

sons and their activity in the city.

Figure 23: The overview visualizations for all persons with

main activity in Milan. From top to bottom: timeline, map,

network graph and collection of pie charts and sunbursts.

REFERENCES

Andr

e, P., Wilson, M. L., Russell, A., Smith, D. A., Owens,

A., and Schraefel, m. (2007). Continuum: Designing

Timelines for Hierarchies, Relationships and Scale. In

Proceedings of the 20th Annual ACM Symposium on

User Interface Software and Technology, UIST ’07,

pages 101–110, New York, NY, USA. ACM.

Borman, T. and Stoel, B. (2009). Review of the uses of com-

puted tomography for analyzing instruments of the vi-

olin family with a focus on the future. J Violin Soc

Am: VSA Papers, 22(1):1–12.

Crauwels, K. and Crauwels, D. (2018). musicmap:

The Genealogy and History of Popular Music Gen-

res from Origin till Present (1870-2016). https://

musicmap.info/ (Accessed 2019-06-24).

den Bulcke, J. V., Loo, D. V., Dierick, M., Masschaele,

B., Hoorebeke, L. V., and Acker, J. V. (2017). Non-

destructive research on wooden musical instruments:

From macro- to microscale imaging with lab-based x-

ray ct systems. Journal of Cultural Heritage, 27:78 –

87. Wooden Musical Instruments Special Issue.

Doi, C. (2017). Connecting music and place: Explor-

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

ing library collection data using geo-visualizations.

Evidence Based Library and Information Practice,

12(2):36–52.

Focht, J. (2006). Bayerisches Musiker-Lexikon Online.

www.bmlo.lmu.de/ (Accessed 2019-12-10).

Focht, J. (2019). musiXplora. www.home.uni-leipzig.de/

mim (Accessed 2019-12-10).

Gleich, M. D., Zhukov, L., and Lang, K. (2005). The world

of music: Sdp layout of high dimensional data. Info

Vis, 2005:100.

Heller, V. (2017). Methoden zur Untersuchung und Doku-

mentation der Geigen am Museum f

ur Musikinstru-

mente der Universit

at Leipzig; Dissertation.

Hopfner, R. (2018). Violinforensic. http:

//www.violinforensic.com (Accessed 2019-06-24).

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. (2015). On Close and Distant Reading in Digi-

tal Humanities: A Survey and Future Challenges. In

Borgo, R., Ganovelli, F., and Viola, I., editors, Eu-

rographics Conference on Visualization (EuroVis) -

STARs. The Eurographics Association.

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

anicke, S. (2018). Visual Ex-

ploration of Musicians and Institutions. Data in Dig-

ital Humanities 2018: Conference Abstracts, 2018

EADH.

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

anicke, S. (2019).

An Interactive Chart of Biography. In 2019 IEEE Pa-

ciﬁc Visualization Symposium (PaciﬁcVis), pages 257–

266.

Khulusi, R., Kusnick, J., Meinecke, C., Gillmann, C.,

Focht, J., and J

anicke, S. (2020). A survey on visual-

izations for musical data. Computer Graphics Forum.

Konopka, D., Schmidt, B., Kaliske, M., and Ehricht, S.

(2017). Structural Assessment of Wooden Musical

Instruments by Simulation: Models, Validation, Ap-

plicability. Proceedings of the 4th Annual Confer-

ence COST FP1302 WoodMusICK - Preservation of

Wooden Musical Instruments Ethics, Practice and As-

sessment.

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

anicke, S. (2020).

A Timeline Metaphor for Analyzing the Relationships

between Musical Instruments and Musical Pieces. In

Proceedings of the 11th International Conference on

Information Visualization Theory and Applications

(IVAPP).

Leskinen, P., Hyv

onen, E., Tuominen, J., et al. (2017). An-

alyzing and visualizing prosopographical linked data

based on biographies. In BD, pages 39–44.

Lu, S. and Akred, J. (2018). History of Rock in 100

Songs. https://svds.com/rockandroll/#thebeatles (Ac-

cessed 2019-06-24).

Meinecke, C. and J

anicke, S. (2018). Visual Analysis

of Engineers’ Biographies and Engineering Branches.

LEVIA18 : Leipzig Symposium on Visualization in

Applications 2018 hrsg. von J

anicke, Stefan; Hotz, In-

grid, 2018.

Miller, M., Walloch, J., and Pattuelli, M. C. (2012). Visual-

izing linked jazz: A web-based tool for social network

analysis and exploration. Proceedings of the Ameri-

can Society for Information Science and Technology,

49(1):1–3.

Munzner, T. (2009). A Nested Model for Visualization De-

sign and Validation. IEEE Transactions on Visualiza-

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

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

Type Taxonomy for Information Visualizations. In

Proceedings., IEEE Symposium on Visual Languages,

pages 336–343. IEEE.

Tresch, J. and Dolan, E. I. (2013). Toward a new organol-

ogy: Instruments of music and science. Osiris,

28(1):278–298.

Tuniz, C., Bernardini, F., Turk, I., Dimkaroski, L., Mancini,

L., and Dreossi, D. (2012). Did neanderthals play mu-

sic? x-ray computed micro-tomography of the divje

babe ﬂute. Archaeometry, 54(3):581–590.

Vavrille, F. (2017). LivePlasma. http:

/www.liveplasma.com/ (Accessed 2019-06-24).

Windhager, F., Federico, P., Schreder, G., Glinka, K., D

ork,

M., Miksch, S., and Mayr, E. (2018). Visualization

of Cultural Heritage Collection Data: State of the Art

and Future Challenges. IEEE Transactions on Visual-

ization and Computer Graphics.

musiXplora: Visual Analysis of a Musicological Encyclopedia