Instructions for the preparation of a camera

Transcription

Instructions for the preparation of a camera
Collaborative annotation of digital images:
the CoDesPho project
Editor(s): Name Surname, University, Country
Solicited review(s): Name Surname, University, Country
Open review(s): Name Surname, University, Country
Valérie Durieuxa,*
a
Department of Information and Communication, Université Libre de Bruxelles,Avenue Franklin D. Roosevelt 50,
1050 Brussels,Belgium
Abstract. Over the last few years, the role of Internet users has progressively shifted from passive consumer of information to
that of an active information producer who interacts with the data. More and more Internet applications such as social bookmarking allow users to create and share metadata describing online resources. Aware of the potential of user-generated metadata to improve access to their collections, some museums and cultural institutions have developed projects that encourage
users‟ engagement with their collection objects. This paper presents the case study of a digital photographs collection from the
Royal Library of Belgium. This amazingly rich collection has never been integrated in the catalogue of the Library because of
the lack of identifying information. In order to help librarians in their descriptive work, the CoDesPho website invites Internet
users to assign comments to photographs. An in-depth analysis of the 392 users' comments describing photographs from the
sample and of the 688 users' queries has been conducted. The content analysis of the comments and their comparison with
users' queries demonstrates that users provide useful and high quality data about the displayed photographs which enable librarians to create a more complete and accurate record for each item.
Keywords: annotation, images description, cultural heritage, user-generated content, queries analysis
1. Introduction
1.1. Collaborative Tagging
In the new version of the Internet commonly referred to as Web 2.0, the role of user has progressively shifted from passive consumer of information to
that of an active information producer who interacts
with the data. More and more Web 2.0 applications
allow these new 'produsers'1 to create and share
knowledge and metadata describing online resources
[2]. This paper focuses on user-generated metadata of
images, and starts with a presentation of the two
creative processes involved: the collaborative tagging
and the social annotation. An analysis of usergenerated metadata, specifically users' comments
regarding online photographs, is then presented. This
study aims to measure the potential of user-generated
metadata to improve access to online photographs.
*
1
The process of ordinary users adding keywords to
online content such as website bookmarks, photographs, computer games or music is called collaborative tagging [7]. The emergent list of terms constituted by all the assigned tags is commonly referred to
as a „folksonomy‟ (short for “folk taxonomy”). It can
thus be defined as a user-generated taxonomy [19].
Collaborative tagging systems can be seen as an alternative method for creating classification systems
[8]. Collaborative tagging and its resulting folksonomies are thus generally compared to professional
indexing and taxonomies. Contrary to the usual tagging classification systems, folksonomies can be described as “anarchic” [9]. Because taggers are al-
Corresponding author. E-mail: [email protected]. Tel. no.: +32486461174
The term 'produser' has been proposed by Bruns to express the fading distinction between consumer and producer of online content.
lowed to assign any word (or even any group of
characters), folksonomies do not have any structure
or hierarchy. Folksonomies therefore do not handle
semantic problems (polysemy, synonymy and basic
level variation) as controlled vocabularies do [10].
Nevertheless, thanks to their lack of structure, folksonomies can evolve fast and readily, to integrate a
neologism, for instance [3]. Moreover, the complete
freedom of the taggers makes collaborative tagging
very easy to perform [14], less time-consuming and
much more cost-effective than professional indexing
[13]. Indeed, the creation and updating of a controlled vocabulary are extremely expensive, and professionals have to be remunerated for indexing the
resources. Finally, according to [12] and [14], using
an uncontrolled vocabulary increases the number of
viewpoints on, and therefore the number of access
points to the tagged content. [16] added that tags assigned by users themselves exactly fulfil their needs;
alternatively indexing performed by experts tries to
guess what their users are thinking. One study in particular conducted at New York‟s Metropolitan Museum of Art has confirmed this assertion. A compari-
son of terms assigned by trained cataloguers and untrained users showed that “non-specialists can supply
a useful number of new access points, augmenting
the professional descriptions of art museum collections” [18].
Aware of the potential of collaborative tagging to
improve access to their collections, museums and
other cultural institutions have developed projects
that encourage users‟ engagement with their collection objects. One of them is “Steve: The Museum
Social Tagging Project” (http://steve.museum/)
which was launched in 2005 by museum professionals. The project aims at addressing concerns by art
museums about access to their ever-growing online
collections. The project team has developed specific
open source software tools for tagging collections
and managing tags, and helps museums implement
them on their own websites or in their galleries. In
the framework of the ongoing project Steve in Action,
a website (http://tagger.steve.museum/) brings together art works from many of Steve's project partners and allows logged users to tag them (see Figure
1).
Fig. 1. Steve in Action Website
1.2. Social Annotation
The access to collection objects can also be improved by another type of user-generated metadata,
which is social annotation. While the collaborative
tagging process results in a chain of one-word de-
scriptors, social annotation systems allow users to
assign free-text comments. As no restrictions are
placed on these comments, they can offer a higher
semantic value for an object‟s description [20].
Since 2005, the National Archives of the Netherlands offers users the possibility of annotating im-
ages from their large collection. The image database
mostly consists of photographs from the former press
agency Anefo that illustrates among other things the
daily life activities from 1945 until 1989. On the detail view of images, a form entitled 'Uw reactie' allows users to write free-text comment (see Figure 2).
Except from a 1,000 character length restriction, no
rules regarding the content or the style of the com-
ment are laid down. In order to avoid spam and inappropriate messages, the comments are reviewed before publication on the website. Once reviewed, the
comments are displayed under the appropriate photograph. All the comments provided by users can be
consulted via a link 'Reacties' on the home-page.
Fig. 2. National Archives of the Nederlands Website
1.3. CoDesPho project
The Royal Library Albert 1st is the national scientific library of Belgium. Its mission is to sustain and
preserve the cultural heritage of the country, and to
make its resources available to the public. Among
these resources is a magnificent collection of digitized photographs dated from 1890 until 1960. They
mostly illustrate the architectural heritage of Brussels,
but also folkloric events, portraits, cities and villages
of Belgium from that time. As the description of the
photographs is incomplete or even missing for most
of them, the collection is not available to the public.
The library does not have the in-house knowledge,
the time, or the money to create descriptions for such
a collection. The emergent phenomenon of usergenerated metadata has thus been envisioned as an
efficient means to collect the missing data. The
CoDesPho project (Collaboration pour la Description des Photographies) was launched in 2007 for
that purpose. A website displaying a sample of the
collection photographs was created. From the homepage, users can access photographs by searching or
browsing the collection (see Figure 3). Two browse
options are available:
 browsing by category - allows users to view
all the photographs belonging to one of the
following
categories:
“architecture”,
“monuments”, “portraits”, “landscapes”,
“daily life and events”, “religious buildings
and pieces of art”; and
 randomized gallery – displays twelve photographs randomly selected in the collection.
From the detail view, which displays the selected
photograph, its description (if existing) and any
posted comments, user can click on a 'Laisser un
commentaire' link which directs them to a form on
which free-text comments can be written and sent
(see Figure 4). The user can fill in name and email
address but this information is not required for posting. Originally, comments were published without
any reviewing process. But with the exponential increase of spamming, they now have to be approved
by a reviewer before publication on the website. A
link on the home-page allows users to consult all the
previously posted comments (see Figure 5). As soon
as the CoDesPho project was initiated, it was highly
successful; during the first month, 392 relevant
comments were posted by users.
In the following section, the relevance of the users
comments posted on the CoDesPho website is analysed along with their potential to improve access to
the annotated photographs.
Fig. 3. Homepage of the CoDesPho Website
Fig. 4. Form allowing users to post comments
Fig. 5. Page displaying all the previously posted comments
graph, and the associated comments, users' name and
posting time) was collected from the CoDesPho website between February 13, 2007 -launching day of the
website- and March 13, 2007. During this first month,
392 relevant comments were posted by users. The
term 'relevant' refers to a provided comment which
content clearly implies connection with the photograph.
Without knowledge to evaluate the accuracy of the
users‟ comments, a 'fitness for purpose' information
quality definition was adopted. The quality criterion
for users' comments is thus fulfilment of the users'
needs. To determine these needs, we collected the
search terms used by internet users when searching
the collection between February 13, 2007 and March
13; this represents 688 terms.
For the content analysis of the users‟ comments
and their comparison with search terms, the Shatford
faceted classification (see Table 1) was adopted. To
assist indexers in their description work of pictures,
Shatford proposes a classification system that divides
the subject of a picture into four attributes: the 'who',
the 'what', the 'where', the 'when' [15]. She draws
these distinctions at each level of meaning identified
by Panofsky in art works: (1) pre-iconographic level
is the primary or natural subject matter (i.e. identifiable objects, people or events); (2) iconographic level
refers to the secondary or conventional subject matter
(i.e. interpretation of actions or gesture); (3) iconological level is the intrinsic meaning or content (i.e.
interpretation of the image based on knowledge and
erudition) [5, 17]. As previous studies have adopted
the Shatford classification, it will allow for comparison of these results with previous work.
2. Related Studies
An in-depth analysis of the users' comments from
the National Archives of the Netherlands has been
conducted by Van Hooland. The study showed that
most comments provide highly relevant information
regarding the photograph they annotate. A mapping
of user's comments with user's queries demonstrated
that “the comments help to fulfil a concrete information need from the user community” [20].
The analysis of the users' query to identify their information needs is not new. Since the eighties, information specialists in visual arts management have
focused on the end-users. In [6], Enser showed that
69% of the users' queries refer to unique entities, be
they in the form of people, objects, locations or
events. It can be explained by the fact that it is less
easy to find the relevant keyword when searching for
“abstract” things than for “unique” entities. It also
established that most queries are subject of refinement in terms of time (the most used), location or
technical specification. [4], [5] and [11] confirmed
Enser's assertions. Batley completed the previous
analyses by focussing on searching behaviour. The
study demonstrated that keywords are mostly used in
searching specific entities whereas random browsing
is used to find abstract (e.g. a busy street) or subjective (e.g. a pretty scene) entities [1].
3. Materials and Methods
This study examines the users‟ comments posted
on the CoDesPho website and tries to measure their
potential to improve access to the annotated collection. Comment data (i.e., the identifier for the photoTable 1
Shatford faceted classification
Pre-iconography
(Generics)
Iconography (Specifics)
Iconology (Abstracts)
Who?
Kind of person or thing
(G1)
Individually named person, group, thing (S1)
Mythical or fictious being
(A1)
What?
Kind of event, action, condition (G2)
Individually name event,
action (S2)
Emotion or abstraction (A2)
Where?
Kind of place: geographical,
architectural (G3)
Individually named geographical location (S3)
Place symbolized (A3)
When?
Cyclical time, season, time
of day (G4)
Linear time: date or period (S4)
Emotion, abstraction symbolized by time (A4)
To make the categories from the Shatford classification more understandable, users‟ comments from
the data set illustrate some of them (see Figure 6).
(G1)
Commentaire envoyé par van assche
jean le 2007-02-22.
Marchandes de lait
(G2)
Commentaire envoyé par Steyaert
Jacques le 2007-03-06.
défilé des écoles. Le détail des médaillons permettrait d'identifier
(G3)
Commentaire envoyé par le 2007-0306-06.
Exact aussi pour la synagogue
(G4)
Commentaire envoyé par Steyaert
Jacques le 2007-03-04.
L'officier de marine porte une casquette
sombre, donc c'est en hiver. En été, on
l'a couvrait de blanc (à cette époque).
(S1)
Commentaire envoyé par Flamencourt,
Marie-Anne le 2007-02-22.
je ne peux identifier le lieu , mais il
s'agit du Roi Baudouin en compagnie
du cardinal Van Roey
(S2)
Commentaire envoyé par Francois Van
Kerckhoven le 2007-02-23.
Expo 58 - Bezoek Prins Albert aan
paviljoen van het Vatikaan ook wel
Cevitas Dei genoemd.
(S3)
Commentaire envoyé par Paul
J.JACQMIN le 2007-02-24.
Pour moi, c'est le coin du bouluvard
Anspach et de la rue du Borgval.
(S4)
Commentaire envoyé par geeraerts
francis le 2007-03-05.
l'année de la photo est 1949
(A1)
(A2)
Commentaire envoyé par jean heyblom
le 2007-02-28.
la statue n'est pas une reproduction de
celle de "La liberté" comme le pensent
beaucoup mais représente un groupe
sculpté par J.De Haan figurant le Progrès entre l'Abondance et la Paix (voir
Le Patrimoine Bruxelles 1 A p 374)
Commentaire envoyé par Josela le
2007-02-26.
Le décor " trompe l'oeil" est amusant ,photos mise en scène de laitières ?? cruches et oeufs dans le panier
Fig. 6. Example of comments belonging to a particular category
from the Shatford classification
4. Results
4.1. Categorizing users' comments within the
Shatford classification
As the users' comments are quiet long (more than
70% of the comments have at least 20 words), they
often belong to several categories of the Shatford
classification (see Figure 7 and 8). Actually 56% of
the comments incorporate more than one category.
Commentaire envoyé par Désiré Roegiest le
2007-03-04.
Il s'agit de la commémoration du 50e anniversaire de la disparition du cardinal Mercier, en 1976, sur le place Ste-Gudule à
Bruxelles, par SM le roi Baudouin.
Fig. 7. Example of a comment belonging to several categories (S1,
S2, S3 and S4)
Commentaire envoyé par mouraux le 200703-03.
En 1954, il y avait un laitier avec une charette tirée par des chiens qui passait regulierement rue Rodenbach a Forest
Fig. 8. Example of a comment belonging to several categories (S3,
S4 and G1)
The mapping (see Table 2) confirms the results of
Enser's and Van Hooland's studies which showed that
users are more interested in unique entities, (or 'specifics' as qualified by Shatford). CoDesPho users
mostly comment on people, objects and locations.
Few comments refer to time period or era.
Table 2
Mapping of users‟ queries to the Shatford classification
Preiconography
(Generics)
Iconography
(Specifics)
Iconology
(Abstracts)
Who ?
14.79% (G1)
55.61% (S1)
0.51% (A1)
What ?
2.04% (G2)
7.40% (S2)
1% (A2)
Where ?
1% (G3)
61.48% (S3)
0% (A3)
When ?
1.02% (G4)
24.23% (S4)
0% (A4)
At the pre-iconographic level, „non-unique‟ people
or objects slightly stands out (G1). Nevertheless, an
in-depth analysis of the comments shows that most of
these comments refer to objects rather than people.
Users tend to identify and name people specifically.
The low percentages at the iconological level
demonstrate that the abstraction is infrequently expressed by users in their comments. The emotions
and abstractions (A2) are usually the expression of an
opinion or a judgement but are quite rare in the data
set (4.85%).
4.2. Users’ comments typology
A further analysis of the users' comments gave rise
to a typology reflecting their most recurrent characteristics. This typology is an adaptation of Van
Hooland's typology to the present data set. As with
the Shatford classification, users' comments can belong to more than one category. From the in-depth
analysis of the users' comment the following categories emerged:
 comments including narrative elements regarding the photographs: 46.43%
 strictly descriptive comments: 26.02%
 comments correcting or confirming the displayed metadata or previous comments:
25.25%
 comments engaging a dialogue with the Library or other users: 19.40%
 comments providing a sound argumentation
(i.e. in-depth analysis of the photograph,
references to books or scientific articles):
10.97%
 comments including user's personal anecdote in relation to the photograph: 5.10%
 comments stating an opinion or a judgement: 4.85%
These categories are neither exclusive nor inclusive, meaning that a comment can belong to more
than one category, but does not necessarily belong to
any.
4.2.1. Comments including narrative elements
Nearly half of the users' comments include narrative elements, meaning they provide further information than what is strictly represented on the photograph. More than just describing the picture, the user
adds information in the form of a narrative. This type
of information generally allows a deeper understanding of the photograph, or even the context of its capture. These narrative comments are quite lengthy.
Some of them amount to more than one hundred
words. The length of the comments demonstrates the
seriousness and strong will of users to help the Library by sharing their knowledge (see Figure 9).
Commentaire envoyé par Yves Randaxhe le
2007-03-01.
Il s'agit en effet de la tour d'angle du bâtiment de
la Banque nationale, à l'angle du bd de Berlaimont et de la rue du Bois Sauvage, dite 'Tour
Beyaert'. A l'intérieur, au sommet de l'escalier,
figurait une statue grandeur nature de l'architecte
Hendrik Beyaert. Cette statue se trouve à présent
dans le hall central du Musée de la Banque
nationale (10 rue du Bois sauvage). Cette tour a
été détruite, avec la plus grande partie du bâtiment Beyaert, lors de la construction du nouveau siège central de la Banque nationale, peu
après la guerre, à l'aplomb de la jonction
ferroviaire Nord-Midi. Voir à ce sujet Walter
Pluym (dir.), L'hôtel du gouverneur de la Banque nationale de Belgique, Pandora, 1995, ou
encore Jozef Victoir e.a., Henri Beyaert. Du
Classicisme à l'Art Nouveau, St. MartensLatem, De Dijle, 1992.
Fig. 9. Example of a comment including narrative elements
4.2.2. Strictly descriptive comments
As opposed to narrative comments, some comments provide information only regarding what is
represented on the photograph (see Figure 10).
Commentaire envoyé par philippon le
2007-02-25.
rue royale au niveau de la colonne du congrès
Fig. 10. Example of a strictly descriptive comment
4.2.3. Correcting/confirming comments
A quarter of the comments in the data set correct
or confirm what is said by previous users (see Figure
11).
Commentaire envoyé par Erik Baptist le
2007-02-21.
C'est belle et bien une ruelle de La Belgique
Joyeuse de l'expo '58 et sûrement pas une
des ruelles de la rue Haute.
Fig. 11. Example of a comment both confirming and correcting
These comments usually respond to the comments
from the fourth category: the ones including a question, but also expressing a doubt by using the conditional mood, for instance (see Figure 12).
Commentaire envoyé par beyens albert
le 2007-02-28.
le personnage avec une croix en dessous
pourrait être le general de gaule avec à
sa droite michel debree?.à gauche du
general? lucien cooremans?.à gauche,de
part et d'autre de notre "flic":deux gendarmes français avec certitude.
Commentaire envoyé par Philippe le 2007-0307.
Cela ne peut pas être l‟Arsenal. Bien possible
par contre que cela soit les casernes Rolin qui se
trouvent de l'autre coté du Bd Gnl Jacques en
direction du bois de la cambre.
Fig. 13. Example of an argument between several users about the
identification of a specific building
Fig. 12. Example of a comment engaging a dialogue
Some of this type of comments fuel arguments regarding the identification of unique entities (see Figure 13).
Commentaire envoyé par claude MEERT le
2007-02-21.
Je crois pouvoir affirmer qu'il s'agit des casernes d'Etterbeek, boulevard Général Jacques.
Commentaire envoyé par Anne le 2007-02-21.
n'est-ce pas plutôt l'arsenal à Etterbeek (au bout
de l'avenue des volontaires - près de la chaussée
de Wavre) les grosses tours carrées du fond sont
occupées par Delvaux et Mer du Nord; les
hangars à l'avant ont disparu; on y construit
actuellement des logements (et bureaux?)
Commentaire envoyé par Gilles le 2007-02-21.
La réponse d'Anne ci-dessous est exacte : il
s'agit bel et bien des anciennes casernes situées
à deux pas de la chaussée de Wavre, au croisement du Bld Général Jacques à Etterbeek. Cette
photo est prise depuis ce qui est actuellement
l'Av des Volontaires. Les bâtiments du fond
existent toujours et abritent effectivement Delvaux et Mer du Nord, le long du Boulevard.
Quant au bâtiment de gauche, je crois que c'est
celui qui abrite actuellement le Colruyt.
Commentaire envoyé par Claude MEERT le
2007-02-22.
J'hésite à donner raison à ceux qui affirment
qu'il s'agit de l'arsenal, parce que les tours occupées par Delvaux et Mer du Nord se situent à
l'avant, très près du Bd Général Jacques et non
dans le fond comme on les voit sur la photo.
Commentaire envoyé par moortgat le 2007-0222.
CASERNE ROLIN
Commentaire envoyé par martine van adorp le
2007-02-23.
c'est bien la caserne Rolin, démolie, qui était en
face de l'arsenal.
Commentaire envoyé par le 2007-03-06.
Il ne s'agit pas, en effet, de l'Arsenal. Celui-ci a
en façade 4 tours à créneaux et sur le coin formé par le bd Général Jacques et la chaussée de
Wavre se trouve une "demi" maison dans le
style de l'époque.
4.2.4. Comments engaging a dialog
Nearly 20% of the comments include a question or
express a doubt by using the conditional mood, for
instance (see Figure 12). As previously shown, these
comments generally lead other users to post confirming comments and thus can be seen as an attempt to
engage a dialog with other users or the Library.
4.2.5. Arguing comments
Some users do much more than provide descriptive information; they perform an in-depth analysis of
the picture and draw attention to details (see Figure
14).
Commentaire envoyé par André Baeck le
2007-03-04.
C'est vrai pour la casquette de marine....
mais les manteaux (qu'on appelait "capote")
sont encore plus caractéristiques de la
mauvaise saison... (je sais bien que le 21
juillet n'est plus un jour de drache nationale
que de beau temps....) Au coude du roi il y
a apparemment un micro sur pied. Il faudrait en déduire qu'il y a eu (ou qu'il y aura)
un discours prononcé depuis la tribune
royale. L'examen de l'exemplaire de base
devrait indiquer si certaines photos sont de
la même cérémonie (format exact, blancheur et rigidité du papier, sans parler
d'indications au verso). Quant aux deux
personnes derrière le Roi, toutes deux en
uniforme sombre avec épée et bicorne
emplumé, on a parlé du plus petit (bourgmestre de Bruxelles ou Grand Maréchal de
la Cour) mais le plus grand devrait aussi
pouvoir être identifié. Quelqu'un dispose-til de collections de journaux des années 50?
Fig. 14. Example of a comment consisting of an in-depth analysis
of the picture
The argumentation can also hinge on references to
books, scientific articles or websites. These references can be very precise (see Figure 15) and are
another demonstration of the motivation and the seriousness of users.
Commentaire envoyé par Steyaert
Jacques le 2007-03-06.
Le personnage de gauche pourrait être
Pierre Vermeylen, prof ULB, ministre
de l'Intérieur ? A droite, c'est bien le
bourgmestre Vandemeulebroeck, voir:
G. H. Dumont, Histoire de Bruxelles,
2005, p 411, Le Cri : accueil de Montgomery.
Fig. 15. Example of a comment providing book references
4.2.6. Anecdotal comments
Only about twenty comments include user's personal anecdotes, meaning information about their
private life or familial history connected with the
photograph (see Figure 16). Whereas these comments
might seem irrelevant, they give to the project a human dimension which certainly positively influences
the motivation and engagement of the users.
Commentaire envoyé par debelder yves
le 2007-02-21.
mes arriére grand parents était marchands de poissons sur ce marché
Fig. 16. Example of a comment providing book references
4.2.7. Comments stating an opinion or a judgement
The least represented category consists of comments expressing an opinion or a judgement of the
user regarding the photograph (see Figure 17). As for
the previous category of comments, their relevance
and utility for the community can be questioned.
Commentaire envoyé par Gui le 200702-21.
Eglise saint Loup à NamurTrèsjoliment
rénové
Fig. 17. Example of a comment illustrating a personal opinion
4.3. Categorizing users' queries within the Shatford
classification
Such as for users' comments, the mapping of users'
queries (see Table 3) confirms the results of Enser's
and Van Hooland's studies which showed that users
are more interested in unique entities. The results
also confirm Batley's theory asserting that users employ keyword search when looking for specific entities, whereas they use browsing for abstract or subjective entities (i.e. a crowded street). Although this
analysis does not confirm that CoDesPho users
browse to find abstract entities, the mapping nevertheless demonstrates that they do not search for them
using keywords queries (see Iconological categories).
Table 3
Mapping of users‟ queries to the Shatford classification
Preiconography
(Generics)
Iconography
(Specifics)
Iconology
(Abstracts)
Who ?
13.66% (G1)
16.13% (S1)
0% (A1)
What ?
0.73% (G2)
2.18% (S2)
0% (A2)
Where ?
1.31% (G3)
58% (S3)
0% (A3)
When ?
0% (G4)
0.30% (S4)
0% (A4)
4.4. Comparing users' comments to users' queries
In order to measure the potential of comments to
fulfil users' needs (that is the quality criterion), a
comparison of the Shatford classification of users'
comments (see Table 2) with users' queries (see Table 3) was made. The mapping demonstrates a general correlation between the content of the comments
and the queries, such as in Van Hooland's study (see
Figure 18).
70
60
50
40
Comments
30
Queries
20
10
0
G1
G2
G3
G4
S1
S2
S3
S4
A1
A2
A3
A4
Fig. 18. Correlation between users‟ comments and users‟ queries
Although most comments and queries belong to
the iconographic level (S), there is a strong difference
between the “who” (S1) and “where” (S3) subcategories. Just as with Van Hooland‟s results, one possibility is that this phenomenon “can be explained by the
interest of users to find images of their home town,
street, or even house, whereas users will mostly
comment upon specific persons, groups, or objects
they know” [20].
A difference is also noticeable between the comments and queries in the “when” subcategory (S4).
While numerous comments provide temporal indications about the pictures, only two queries belong to
this category. This might confirm Enser's results
which showed that time period indications are mostly
used as refiners of queries [6]. As the CoDesPho
website does not allow refinement functionality, it is
impossible to verify this assertion.
The comparison between comments and queries
also allows evaluating the relevance of the comments
stating an opinion or judgement. As previously mentioned, the opinions and judgements are generally
expressed by emotions and abstractions (A2); none of
the queries belongs to this category. Comments stating an opinion or a judgement are useless for the
community.
Despite these differences, the comparison of comments with queries shows strong similarities in their
content. It demonstrates that the comments tend to
meet the information needs of the users.
5. Conclusion
The description of art works, and photographs in
particular, requires a large amount of in-house
knowledge, time and money which museums and
cultural institutions often do not have. Some magnificent collections thus remain inaccessible to the public
because of a lack of descriptive metadata. The emergent phenomenon of user-generated metadata could
constitute an efficient means to collect the missing
data.
This analysis of the CoDesPho users' comments
describing photographs from the Royal Library of
Belgium establishes the seriousness and strong will of
users to help the Library by sharing their knowledge.
The comparison between the users' comments and
queries shows the high quality of the users' annotations. The strong correlation between the comments
and queries indeed demonstrates that the comments
help to fulfil the information need of the users.
Despite the highly positive evaluation of users‟
comments quality, the CoDesPho project has been
discontinued as of June 2007. The hundreds of comments posted on the websites have never been used
by the Royal Library. As of 2010, this magnificent
collection of photographs is still unavailable to the
public.
References
[1] Batley S. Visual information retrieval: browsing strategies in
pictorial databases. Proceedings of the 12th International Online Information Meeting (London, UK, December 6th-8th,
1988). Oxford: Learned Information, 373-381.
[2] Brun A. Blogs, Wikipedia, Second life, and Beyond: from
production to produsage. New York : Peter Lang, 2008.
[3] Cattuto C., Loreto V. and Petronero L. Semiotic Dynamics
and Collaborative tagging. Proceedings of the National Academy of Sciences of the United States of America,
104(5):1461-1464, 2007.
[4] Chen H. An Analysis of Image Queries in the Field of Art
History. Journal of American Society for Information Science
and Technology, 52(3):260-273, 2001.
[5] Choi Y. and Rasmussen E. M. Searching for images: The
analysis of users‟ queries for image retrieval in American history. Journal of the American Society for Information Science
and Technology, 54(6):498-511, 2003.
[6] Enser P.G.B. Query analysis in a visual information retrieval
context. Journal of Document and Text Management, 1(1):2552, 1993.
[7] Golder S.A. and Huberman J.L. Usage Patterns of Collaborative tagging systems. Journal of Information Science,
32(2):198-208, Apr 2006.
[8] Kipp M.E. Complementary or Discrete Contexts in Online
Indexing: A Comparison of User, Creator and Intermediary
Keywords. Canadian Journal of Information and Library Science, 29(4):419-436, 2005.
[9] Lambiotte R. and Ausloos M. Collaborative tagging as a tripartite network. Lecture Notes in Computer Science,
3993:1114-1117, 2006.
[10] Macgregor G. and McCulloch E. Collaborative tagging as a
knowledge organisation and resource discovery tool. Library
Review, 55(5):291-300, 2006.
[11] Markkula and M. Sormunen E. End-User Searching Challenges Indexing Practices in the Digital Newspaper Photo Archive. Information Retrieval, 1:259-285, 2000.
[12] Mathes A. Folksonomies – Cooperative Classification and
Communication Through Shared Metadata. Computer Mediated Communication, December 2004. Retrieved August 15,
2010, from: http://adammathes.com/academic/computermediated-communication/folksonomies-old.html.
[13] Merholz P. Ethnoclassification and vernacular vocabularies.
August 30, 2004. Retrieved August 15, 2010,
from: http://www.peterme.com/archives/000387.html.
[14] Mesnage C. and Jazayeri M. Specifying the collaborative
tagging system. Proceedings of the 1st Semantic Authoring
and Annotation Workshop located at the 5th International Semantic Web Conference (Athens, GA, USA, November 6th,
2006).
[15] Shatford S. Some Issues in the Indexing of Images. Journal of
the American Society for Information Science and Technology,
45(8):583-588, 1994.
[16] Shirky C. Ontology is Overrated: Categories, Links and Tags.
Shirky.com, 2005. Retrieved August 15, 2010, from:
http://shirky.com/writings/ontology_overrated.html.
[17] Svenonius E. Access to Nonbook Materials: The Limits of
Subject Indexing for Visual and Aural Languages. Journal of
the American Society for Information Science and Technonlogy, 45(8):600-606, 1994.
[18] Trant J. Exploring the potential for social tagging and folksonomy in art museum: proof of concept. The New Review of
Hypermedia and Multimedia. 12(1):83-105, 2006.
[19] Vander Wal T. Explaining and Showing Broad and Narrow
Folksonomies. vanderwal.net, 2005. Retrieved August 10,
2010,
from:
http://www.vanderwal.net/random/entrysel.php?blog=1635.
[20] Van Hooland S. From Spectator to Annotator: Possibilities
offered by User-Generated Metadata for Digital Cultural Heriage Collections. Proceedings CILIP Cataloguing & Indexing
Group Annual Conference (Norwich, UK, September 13th15th, 2006)