EDUCATION WITH “LIVING ARTWORKS” IN MUSEUMS

Roberta Beccaceci

(?)

, Francesca Fallucchi

(†)

, Cristina Giannone

(†)

Francesca Spagnoulo

(?)

and Fabio Massimo Zanzotto

(?,†)

(†) Dipartimento di Informatica, Sistemi e Produzione, University of Rome ”Tor Vergata”, Italy

(?) Facolt

a di Lettere e Filosoﬁa, University of Rome ”Tor Vergata”, Italy

Keywords:

Museums, Conversetional Agents.

Abstract:

Museums need to ﬁnd innovative ways of communicating if these institutions want to survive in the new era

and want to play their active role of educators. In this paper, we will present our idea of living artworks. Using

conversational agents we want to give artworks the capability of talking to visitors. A living artwork attracts

attention, being a funny and novel combination of art and technology. The mix of experience and action has a

beneﬁcial effect in learning new concepts or facts. We will then present our methodology for building living

artworks, the enabling technologies, and a case study.

1 INTRODUCTION

There is a growing interest in using museums as

places where to informally deliver educational con-

tents (Severino, 2007). Artworks are not simply

“artists’ messages to audience” but also tools to create

links between cultural contents and audience. Yet, to

positively play the role of informal educators, muse-

ums have to deliver content in a novel way. Accord-

ing to the analysis reported in (De Biase, 2008), many

visitors, around 33%, forget the names of artists and

the artworks they saw into museum, 50% of the visi-

tors remember not much about artworks, and 32% of

them mixed subjects up with others never seen. These

are clear indicators that somehow traditional muse-

ums are not well suited to play this important role of

informal educators.

To win against different media and forms of en-

tertainment, museums are seeking innovative ways to

deliver their contents to the public. In their battle

for surviving, these institutions are opening to novel

technologies. There are active communities look-

ing and monitoring new trends in technologies. The

MuseTech Central

is a place to share information

about technology-related museum projects such as so-

cial tagging for artworks (Trant and Wyman, 2006).

The awareness of the role of informal educators and

the technology fertile ground make museums won-

derful places where to experiment with novel and in-

http://www.musetechcentral.org/

formal ways of communicating educational content.

This will be beneﬁcial both for museums and for au-

diences.

Our idea is to create living artworks. We want to

give to artworks the capability of talking, in order to

have a “dialogue” with visitors. We will use conver-

sational agents. In a nutshell, conversational agents

are software systems that can have conversations with

users. These software systems come from the tradi-

tion of Artiﬁcial Intelligence (AI). Building systems

able to replicate the human ability to communicate is

one of AI’s main goals, see for example the Turing

Test (Turing, 1950). Conversational agents have been

already used for educational purposes and we believe

that artworks that can tell their own history will attract

people.

In this paper, we will present this idea of living

artworks. We shortly review the related work, i.e., the

use of conversational agents as tutoring systems (Sec.

2). We will then present enabling technologies and

our methodology for building living artworks (Sec.

3). We ﬁnally describe the application of our method-

ology (Sec. 4).

2 CONVERSATIONAL AGENTS

IN EDUCATION

Education is showing a growing interest towards new

technologies and new media as innovative means to

346

Beccaceci R., Fallucchi F., Giannone C., Spagnoulo F. and Massimo Zanzotto F. (2009).

EDUCATION WITH “LIVING ARTWORKS” IN MUSEUMS.

In Proceedings of the First International Conference on Computer Supported Education, pages 346-349

DOI: 10.5220/0002158903460349

 SciTePress

support learning. Multimedia technologies mix lan-

guage and images that are the basic blocks in the

process of thinking and learning. These technologies

build new suitable communication channels for teach-

ers that want to exchange educational messages with

their students. Yet, technology will not substitute but

broaden pedagogical strategies available for teachers

(Dede, 1992).

Artiﬁcial Intelligence (AI) has been often seen as

an interesting way to support traditional and distance

learning. AI systems can make distance learning more

attractive (Moore and Kearsley, 2005). People, in

general, and children, in particular, are increasingly

more acquainted with contents delivered with new

technologies. Children play with videogames and

chat over the Internet with astonishing ease. Using

new technological channels can attract students stim-

ulating their curiosity. A simple way to introduce new

technologies is to use software embodied agents, i.e.,

virtual puppets on a screen, as storyteller. In (Dami-

ano et al., 2006), these agents have been used to guide

children inside a museum. A more attractive way is to

use interactive agents called Conversational Agents.

These are able to establish interactive communica-

tion with users using natural language. Conversa-

tional agents have been used for language education

in (Jia and Ruan, 2008) or for teaching physical and

procedural tasks using a virtual reality environment

in (Rickel and Johnson, 1998). Some recent works in

Human-Computer Interaction ﬁeld (HCI) (Core et al.,

2006) explored the use of virtual agents for educa-

tional and training. In this environment, users train

their skills interacting with virtual agents like in a

videogame.

3 REVITALIZING MUSEUMS

WITH “LIVING ARTWORKS”

Museums need to ﬁnd innovative ways of communi-

cating if these institutions want to survive to the new

era and want to play their active role of educators.

We believe that our idea of living artworks, i.e., an

artwork extended with the capability of talking about

itself, can help in this difﬁcult challenge. The inte-

gration of different cultural experiences adds value to

knowledge accumulation process (Severino, 2005).

We ﬁrstly deﬁne a living artwork and analyze the

beneﬁcial effects for museums (Sec. 3.1). We then

present the enabling technology, i.e. the stimulus-

response conversational agents, (Sec. 3.2) and the

procedure to use it (Sec. 3.3).

3.1 “Living artworks” as Educational

Tools in Museums

A living artwork is a combination of an artwork and

a conversational agent. We want to extend the posi-

tive experience of storytelling in museums (Damiano

et al., 2006) integrating principles and technologies of

conversational agents used for educational purposes

(Core et al., 2006). The agent should involve visitors

in conversations about the artwork.

A living artwork wants to attract attention, being

a funny and novel combination of art and technology.

At the same time, it wants to generate new knowledge.

It acts as instrument of cultural communication and as

tool of cultural production. A living artwork allows

a bidirectional communication helping the shift from

passive spectator to active spectator. The combina-

tion of experience with action has a beneﬁcial effect

in learning new concepts or facts.

Given an artwork, we can build the related living

artwork if:

• we have a conversational agent platform with a

simple language for describing the knowledge

needed for the dialogue

• we completely deﬁne the knowledge base to have

conversations about the artwork

In the next sections, we ﬁrstly introduce the conversa-

tional agent platform we use (Sec. 3.2) and secondly

we deﬁne a methodology to build the knowledge base

(Sec. 3.3).

3.2 Conversational Agent Technology

A Conversational Agent (CA) is a software tech-

nology that uses the natural language utterances to

interact with users. Conversational agents can “have

a conversation with a user”. Among all the possible

CA technologies, we selected the simplest, i.e.,

those following the Behavioral psychological theory

(Watson, 1928). According to this theory, human

mind can be studied only relating received stimulus

and emitted responses. The human behaviors can be

described by the following function:

R = f (S) (1)

When an individual is exposed to stimuli his response

is a function of these stimuli.

The behaviorism has inspired one of the ﬁrst em-

ulative conversational agents. In 1966 Weizenbaum

build a software program called Eliza (Weizenbaum,

1966). This artiﬁcial agent simulated the Rogersian

psychologist behavior. The interactions was grounded

on the agent ability to hold control about dialogue,

EDUCATION WITH "LIVING ARTWORKS" IN MUSEUMS

347

using as ﬁrst interaction (stimulus) a question for the

user, which will response based on stimulus and his

environment. After this, the interest on conversa-

tional agent ﬁeld grew up, drawing the attention of

many different research ﬁelds like Psycholinguistics

or Software engineering. An important mention for

ALICE (Artiﬁcial Linguistic Internet Computer En-

tity) (Wallace, 2004) that implements a simple frame-

work allows somebody to create its conversational

agent by creating a collection of stimulus-response

pairs. When a user will ask a question to the agent,

the stimulus is searched in the collection. If this stim-

ulus is in the collection, the agent will reply with the

related response.

In this study we use ALICE as it offers a simple

language to write stimulus-response patterns. This

simple technology in the last years has inspired many

research works that proposed extension of this tech-

nology (e.g. (Pilato et al., 2004)).

3.3 Building Personality of

Conversational Agents

As we saw in the previous section, ALICE is simple to

use. Building the agent personality, i.e. its way to an-

swer to questions, means to write stimulus-response

pairs. Furthermore this technology allows to model

and store data about current dialogue topic, other data

about user, or dialogue in general. This makes possi-

ble to program the agent to reply to a great amount of

questions with different languages, e.g., a colloquial

language with idioms or jokes.

The process we follow to build the collection of

stimulus-response pairs has two phases:

• Wizard-of-OZ phase: using a standard model to

collect data for dialogue systems, i.e., the Wizard-

of-OZ (Fraser and Gilbert, 1991), our ﬁrst phase

is to collect possible dialogues using people that

simulate the behavior of the system. Dialogues

collected in this phase are used to produce a ﬁrst

collection of stimulus-response pairs.

• Controlled Wizard-of-OZ phase: the system is de-

ployed with the ﬁrst collection of stimulus re-

sponses. Dialogues are logged in order to analyze

failures. The initial collection is then augmented

with new stimulus response covering failures.

4 CASE STUDY

In this section we present an initial application of the

living artwork concept. We describe the target audi-

ence and the target artwork. We then report on the

related personality-building phase.

4.1 Target Audience and Target

Artwork

Children with an age range 8-12 years are our initial

target audience. We want to try if it is possible to cap-

ture their attention on artworks for which they do not

show a natural interest. We believe that children can

be more open to technological communication chan-

nels and that a novel way of approaching artwork with

entertaining conversational agents will help them in

better learning.

Among many artworks, we selected a painting

representing a scene with many people having in-

teresting story to tell. Many subjects in the scene

give us the possibility of better experimenting with a

conversational technology as we can focus the atten-

tion in many parts of the painting. We then selected

the painting “The Hunt of Diana” (see Figure 1),

placed in Rome in the Borghese Gallery and Museum,

painted by Domenico Zampieri called Domenichino

in 1616/17. In his masterpiece Domenichino revived

antique literary and mythological themes such as an

archery scene deriving from Virgil’s Aeneid (V,485)

and the depiction of nymphs. Many characters in the

scene have a story to tell.

We divided the painting in ﬁve areas (see Fig-

ure 1). Each one connected to a particular subject,

according to the general intention to let audience to

choice the topic to study in depth.

Figure 1: “The Hunt of Diana” painting.

4.2 Building the Conversational Agent

Personality for the Case Study

For the construction of the personality, we followed

the two-step procedure described in Sec. 3.3: the

ﬁrst phase where the system is simulated, the Wizard-

of-Oz phase, and the second phase where the system

is deployed and its failures detected. Before starting

CSEDU 2009 - International Conference on Computer Supported Education

348

Table 1: A sample dialogue with the simulated Agent.

Agent: Hello, do you want to know something about this

painting? Play with us and answer to some ques-

tions. What is your name?

Child: asia

Agent: Hello Asia, do you know the title of the painting ?

Child: no

Agent: It’s ”The Hunt of Diana” and it placed in Rome in

the Borghese Gallery and Museum. It was painted

by Domenico Zampieri in 1616/17. Do you know

why it is called Domenichino (Little sunday)?

Child: because he works only on sunday.

Agent: Clever but it is not the right answer. He was short!

...

we collected a document describing the particular art-

work. We organized this document as if it were a

collection of stimulus-response. It contains general

information about painting and painter’s life, infor-

mation about speciﬁc parts of the painting, and par-

ticularities of the artwork. We used this document to

drive the ﬁrst phase.

We organized the Wizard-of-Oz phase as follows.

We used 3 classes in a primary school as focus group.

The Conversational Agent has been simulated with

Microsoft MSN. The language used by the simulated

system, i.e. the operator, is simple and colloquial. At

the end of this phase, we gathered 10 dialogues. An

example of these dialogues can be seen in Tab. 1.

This set of dialogues has been used to design the real

stimulus-response collection. We deﬁned around 130

stimulus-response pairs.

The second phase is currently running. The sys-

tem has been deployed on a web site and it is accessi-

ble over the net

5 CONCLUSIONS AND FUTURE

WORK

This paper presents a novel idea that we called “liv-

ing artworks”. We believe that this paradigm can be

useful in two ways: making museums more attractive

places and increasing the effectiveness of museums

as informal educators. This is a preliminary work and

we need to study if the above two claims can be sup-

ported with empirical evidence. We then need to de-

ploy our “living artworks” in a museum and measure

whether the knowledge retention indicators (as those

used in (De Biase, 2008)) have a beneﬁcial effect.

http://lirﬁ.lettere.uniroma2.it/ArtiD

REFERENCES

Core, M., Traum, D., Lane, H. C., Swartout, W., Gratch,

J., Lent, M. V., and Marsella, S. (2006). Teaching

negotiation skills through practice and reﬂection with

virtual humans. Simulation, 82(11):685–701.

Damiano, R., Lombardo, V., Nunnari, F., and Pizzo, A.

(2006). Dramatization meets narrative presentations.

In Mobile Guide 06.

De Biase, F. (2008). La centralita’ dello spettatore. In

L’arte dello spettatore, il pubblico della cultura, Mi-

lano. Franco Angeli.

Dede, C. (1992). Education in the 21st century. In Annals

of the American Academy for Political and Social Sci-

ence, pages 104–115.

Fraser, N. and Gilbert, G. N. (1991). Simulating speech

systems. Computer, Speech and Language, (5):81–99.

Jia, J. and Ruan, M. (2008). Use chatbot csiec to facili-

tate the individual learning in english instruction: A

case study. In Woolf, B. P., Ameur, E., Nkambou, R.,

and Lajoie, S. P., editors, Intelligent Tutoring Systems,

volume 5091 of Lecture Notes in Computer Science,

pages 706–708. Springer.

Moore, M. and Kearsley, G. (2005). Distance Education: A

Systems View. Thomson/Wadsworth.

Pilato, G., Vassallo, G., Augello, A., Vasile, M., and Gaglio,

S. (2004). Expert chat-bots for cultural heritage. In

IX Convegno della Associazione Italiana Intelligenza

Artiﬁciale Proc. of. Workshop Interazione e Comuni-

cazione Visuale nei Beni Culturali, page 15.

Rickel, J. and Johnson, W. L. (1998). Task-oriented dialogs

with animated agents in virtual reality. In Workshop on

Embodied Conversational Characters. In Proceedings

of the Workshop on Embodied Conversational Char-

acters, pages 39–46.

Severino, F. (2005). Un marketing per la cultura. Franco

Angeli, Milano.

Severino, F. (2007). Comunicare la cultura. Franco Angeli,

Milano.

Trant, J. and Wyman, B. (2006). Investigating so-

cial tagging and folksonomy in art museums with

steve.museum. In WWW 2006.

Turing, A. (1950). Computing machinery and intelligence.

Mind, 59:433–460.

Wallace, R. (2004). The elements of aiml style. alice ai

foundation.

Watson, J. B. (1928). The ways of behaviorism. Commun.

ACM.

Weizenbaum, J. (1966). Eliza—a computer program for

the study of natural language communication between

man and machine. Commun. ACM, 9(1):36–45.

EDUCATION WITH "LIVING ARTWORKS" IN MUSEUMS

349