|
 Ewan Andreasen |
Abstract
The thesis concerns itself with the design of a thesaurus for use with the indexing
of press photos. The paper begins with theories regarding picture analysis and
levels of meaning in subject analysis of press photos. Empirical analysis of a
domain context for press photos is made on site at a Danish regional newspaper.
Data from these two approaches are then merged for use with the formulation of
indexing rules for the indexing of press photos plus principles for the design
of a thesaurus supporting these. The author concludes that thesaurus design for
best results must begin with an analysis of that domain which the thesaurus
will be operating in. During the analyses a method was developed for handling
index terms containing complex meaning, called the Visual Prototype Program.
Danish Library school student Ewan Andreasen submitted this thesis for his masters degree (Cand.Scient.Bibl.) at the Danmarks Biblioteksskole in Aalborg, Denmark, in May 2002. During his research, he accessed The Caldeson Consultancy Guide to Picture Concept Indexing and used its ideas as a basis for some of his theories on image index in a news media environment. He cites the Caldeson work on page 34 of his thesis (page 37 of the on-line file).
III.4: Is it flying or jumping? |
He writes: “You might be interested some day to read the work by Russell & Fernandez-Dols (1997), see my references. My own theoretical contribution was to assume that the meaning level in pictures consists of a descriptive and a derivative level. As I write on (page 32 of the .pdf file), with the image of this four-wheeled vehicle, you cannot discern whether it really is flying, floating, falling, jumping, etc. Thus, the descriptive characteristics for all these derivative index terms are the same.
“They are derivative, since they derive some of their meaning from the descriptive level. Now, should the vehicle’s action be index with the keyword ‘jumping’, this will be a product of the indexer’s own understanding, whether this understanding be based on subjective feelings or cultural conventions.”
He quotes U.S. academic Brian Stonehill, researcher in on-line visual literacy (OVL) at the Pomona College in Claremont, California, who wrote in a 1998 report on the OVL project: ”Motion in visual images is always an illusion. It can either be implied or suggested, as in still pictures... recognizing motion in still images is a more conscious, interpretive process.”
Andreasen contends that the keyword “jumping” would most properly be used for indexing his quad-bike picture but that indexers could, if they wished, choose to write a caption that made the reader or viewer see that the sand buggy was, for example, flying not jumping. The thinking parallels the Caldeson model in that one image could convey different meanings all at the same time.
The thesis is in the Danish language. The Records and Information Management On-line Service (RIMOS) plans to present it completely in English in the near future. The author has translated a considerable portion of his work into English text which begins below. The original thesis, Indeksord for Pressefoto: Designprincipper for en tesaurus, comprises illustrations and some 30,000 words making an Acrobat Adobe 4.0 .PDF file of a little over one megabyte. Downloading can take up to six minutes.
 |
Why a thesaurus? A thesaurus is still necessary today because it remains impossible to organize knowledge by computer technology alone. The efforts of indexing and retrieval of digital images, for example, can be divided into two categories: Content- and concept-based (see Chen & Rasmussen, 1999, and Jörgensen et al., 2001). The content-based efforts involve computer analyses of images at the pixel level, and because of this they only deal with characteristics like colour composition, dots and lines, combinations of these and so on.
However, content-based solutions are cheap and fast. An image is analysed at a very deep level and in incredible detail, but the resulting syntactical data does not necessarily equal the pieces of information that the user might need during retrieval. The need for semantic information (which is characteristic for content-based efforts) still exists. The problem with these is the complexity introduced along with the definition, construction and maintenance of such systems and this point will be emphasized also in this text. The manual indexing of images is an expensive method and can introduce inter-indexer inconsistency. Still, this is an acceptable loss compared to the semantic vacuum which the content-based efforts leave behind.
Construction of thesauri whose terms are used for the indexing of images is done under conditions that are different from the construction of thesauri whose vocabulary totally or partly can be extracted from text-based references (such as documents). More specifically, this paper evolves around the development of theories of meaning and visual understanding of Press photographs. The theories are supported empirically by data collection at a Danish regional newspaper. Theory and empirical data is then used for the development of basic design principles for a thesaurus for use in the indexing of Press photos.
According to the problem definition, this paper has used subject thesauri as research for indexing visual meanings which have no textual dimension. These can be static or dynamic images, but this text focuses on static images. Since dynamic images change visually over time, they are assumed to require a different approach to indexing from static ones. Furthermore, static images could be categorized according to type. Here, the photograph is chosen rather than paintings or drawings because of its ability to depict reality with extreme accuracy versus replicating it in an abstract or interpretive manner.
But the photograph as an image type must be further subdivided. Photos can be manipulated and can therefore enter the artistic or abstract spheres. I want to distinguish between the types: art, advertising and Press photographs. Hatano (Ørnager, 1996) identifies four types, as he also mentions the documentary photo. Here, I have chosen to see Press photographs and documentary photographs as representing one type. The typology is based on several elements, including the visual expression, the modality of the photograph as presented by Kress & Leeuwen (1996, p. 159) as well as the symbiosis between the photograph and the caption text, which is more than just descriptive data. Caption text is a small mass of text that places the photograph in a context.
More specific definitions of the concepts “modality (“mood”)"and “caption text” are appropriate before I characterize the three photograph types:
The expression has its origin in linguistics and means ”truth value”, and is here used similarly in the study of images. In this context, truth is not final, but includes different opinions about the notion of “truth”. Thus, a photograph's depiction of reality is only one possible definition of it. What separates the photograph from paintings, drawings and diagrams is our (the observers) perception of its modality. We have gotten used to the standards of photography, 35mm colour, and we generally believe that photographs never lie. Kress & Leeuwen describe several forms of modality (for different types of images, see Kress & Leeuwen, p. 170), where they cite Bernstein. An example of this is to be found in the scientific domain, where black and white diagrams with a maximum level of abstraction are viewed to have the greatest amount of modality.
Therefore “modality” is part of a social-semiotic theory of truth. No one can say that something is true, only that it is represented as true. The social dimension plays a part in that we as viewers think our culture into the interpretation of , for example, the photograph:
"...det er med min kultur [...] jeg deltager i disse figurer, ansigtsudtryk, gestus, omgivelser og handlinger." (Barthes, 1996).
Photographs are often seen as having high modality, and that is partly due to the incredible amount of detail which was introduced with the advent of photo-technology. Suddenly reality could be reproduced analogically. Barthes mentions that the photo is random (“…det gentager mekanisk det som aldrig mere vil kunne gentages eksistentielt...", Barthes 1996, p. 12) as well as invisible ("...ligegyldigt hvad det viser eller hvordan det gør det, så er det ikke fotoet man ser.", Barthes 1996, p. 15), and this has contributed to create our great belief in its depictions.
Caption text is text that appears in the immediate vicinity of an image. Barthes (1977) has said that the meaning of a photograph is always dependent on verbal text. Photographs in themselves are simply too ambiguous, and transfer an unlimited stream of visual meanings. Together with the text, however, one of these meanings is emphasized, either via illustration, in which the text is “read” before the picture, or anchoring, in which the text is “read” after the image, and serves to define its meaning. Admittedly, it is common that in newspapers and magazines we see pictures in connection with a caption text. It is to be expected that the caption text influences our interpretation of these images to a great extent.
Artistic photographs: This type can be naturalistic, but often do not emphasize the modality of the image. Rather, they concentrate on a message or an experience at the expense of its mood. Artistic photographs display a lot of experimentation with visual expression and the visual elements. Typically, it is obvious that the photograph and its version of reality have been tampered with, either by colour saturation, photographic effects, strange perspective or the like. Artistic photographs are often without a caption text, and only exist with the descriptive data assigned to it (year, the identity of the photographer, title). Furthermore, the title of an artistic photo (if present) can sometimes appear mystic or misleading, according to the intentions and message of the photographer.
Commercial photographs: These do not display a particular high modality, either, but seek to emphasize the emotions they can arouse in the viewer. Photographic effects and means are used explicitly to acquire a positive response in the viewer towards a product or service which is the subject of the picture. They can actually consciously violate the conventions of the truth value of the photo. For example, colours may appear more saturated than they are in reality. Commercial photographs are, like many artistic images, arranged or constructed. The caption text on a commercial photograph is used to consolidate the product or service.
Press photographs: In the case of Press pictures, it is implicitly important to maintain the truth value of the image, not forgetting that Press photographs also can be subject to manipulations perhaps to display particular political values, for example. This manifests itself because journalists consciously or subconsciously operate under certain conventions in their acts of illustrating and explaining reality. Where commercial photographs have as a goal the creation of a positive disposition towards a product or service, the goal of the Press picture is to give a feeling of authenticity, a sense that what is being displayed really is or has been. It can be argued whether the way it is expressed represents the truth, but nonetheless the Press picture is seen to have a high modality. In this paper, I limit the object of study to the Press photograph. One of the reasons is that I see journalists as participants in a certain discourse, one that can be illuminated in order to influence the construction of a thesaurus. Another reason is that, in its effort to observe a certain modality, the Press photograph creates a ”sphere of stability” that makes it especially suitable as an object for indexing. The expression ”naturalistic coding” expresses the modality of the Press photograph (see Kress & Leeuwen, p. 171). We are used to observe ”natural” colours in a Press picture, and we would view over-saturated colours as an unacceptable manipulation of the image. Strange perspective is sometimes seen, but in general we like Press photographs to be displayed without any distortion or digital manipulation (no matter that it is typically unnoticed when it is done). One can say that the viewer and the photographer have entered into an unwritten agreement, and I am going to assume that this ”contract” is a part of that sphere of discourse in which the Press is operating. Now that the Press photograph has been identified as the bulk of
referents, which an hypothetical information system is to organize, we examine
the second focus of the problem: The construction of a thesaurus for indexing
Press images. As a framework for this section, Batty (1989) is used as
inspiration. In Batty's approach, acquiring the base vocabulary comes before
the analysis of facets. Another approach could be to define facets before
acquiring the vocabulary. This will be decided in a later chapter. Thus, the
construction of a thesaurus is seen as consisting of: Thesaurus definition: Lykke Nielsen (2001) has made a framework for thesaurus design based on
the context of the information system. Her approach to defining and designing a
thesaurus focused on domain analysis as an important source of data. Her domain
analysis consists of the following elements: In this paper, a domain analysis we focus analysis on the journalistic
domain, whose members produce and use Press photographs. The domain analysis
uses, as a starting point, the theories of journalism. The regional newspaper Nordjyske
Stiftstidende (NS) was chosen to illustrate empirically the
theoretical issues. Interviews and association test were carried out in the
editorial offices, the natural environment of the workers. Group interviews were carried out during April 2002 with employees
from the NS. These interviews took the form of an open, qualitative
structure. For a satisfactory result, three groups were interviewed: Photographers,
representing the producers of the research object; the librarians in the
Picture Library as the information intermediaries and, finally, the
journalists, representing the users of the research object. Analysis of user requests was carried out through
interviews with the picture librarians. Unfortunately, no records of user
requests from the past were available. The intermediary function was usually
carried out over the phone or in direct face-to-face communication. Word association tests were performed using Press photographs as
stimuli. A wide variety of journalists carried out the test. It is arguable to
what extent this method contributed to an understanding of the terminology in
use by journalists, specifically in connection with Press pictures.
Furthermore, an alternative test, in which participants identify images from
words is considered, possibly uncovering other aspects of vocabulary use. Thus, the first research question is formulated as follows: Research question 1: How does an analysis of the context of the
Press photograph, defined as the journalistic discourse, contribute to defining
a thesaurus structure? Gathering the vocabulary and facet analysis: The problem with gathering a vocabulary in the domain of Press
photographs is, of course, that images do not communicate their content in a
verbal language. Normally, index terms are gathered in such a way as to have
both literary or user warrant (Lancaster 1986, p. 28). It is usually best to
use both these approaches, but the photograph as a source for terms leaves only
user warrant as a possibility. The basic structure of the thesaurus must be theoretically
founded, and study of the Press photograph in its role as a carrier of meaning
helps define this structure. The vocabulary and the fundamental facets will thus be based on the
photograph itself as well as the journalistic discourse. The next preliminary
research questions can be formulated as follows: Research question 2: How can the vocabulary of the thesaurus be
based on the journalistic discourse? Research question 3: In which way is the internal structure of the
thesaurus affected by the information object, the Press photograph, whose
indexing it must support? Use of thesaurus: A thesaurus is not completed without instruction in its use. This
involves an analysis of the subjects contained in a Press photograph. Indexing
images can be difficult, and therefore, parts of this paper are concerned with
indexing Press photographs. It deduces some practical principles to this end.
These principles are, like the design of the thesaurus itself, based on an
understanding of the discourse at the newspaper or journalistic discussion as
such. The next research question can be defined as follows: Research question 4: What principles can be deduced for indexing Press photographs, based on a) the journalistic discourse and b) the visual structure of the Press photograph? Used literature: I attempt to reach a component-based approach to the Press image, based
on an image-analysis approach by Kress & Leeuwen (1996). Their primary idea
is that visual material can be viewed as having a syntactical structure.
Through syntactical analysis and division of the picture into components, a
greater understanding of the image as a whole is reached. I seek to use this
approach in photograph indexing, so that it might parallel the way text
documents can be split up into component parts to assist indexing. Their work
is concerned with images of all types, and can be used for analysing Press
photographs. I have chosen to introduce my work by condensing and explaining
Kress & Leeuwen's approach since it provides me with suitable terminology
that I will use in the chapters to come. However, the Dutchmen do not contribute an understanding of the deeper
levels of meaning in images. This is provided primarily by Roland Barthes, a
French semiologist who has been working closely on the Press photograph and its
meaning. Sara Shatford (1986) has contributed with an approach to subjects in
images founded on information science, in which she draws upon Panofsky in
particular. In this paper, my approach to subject analysis is based upon a
synthesis of these two. Previous research: Writing this paper, I have identified a Danish PhD. thesis from Susanne
Ørnager (1999), who also has been working with indexing and retrieval of Press
photographs. Her approach is based on previous work of hers, which includes a
thorough analysis of work practice as it appears in picture libraries in Danish
newspapers. Her research reaches some of the same conclusions that I have
regarding the meaning levels of Press photographs. She has also included the
theories of Roland Barthes in her work.
As in the approach of Lykke Nielsen described above, Ørnager has used
word-association tests in her analyses. As mentioned, I try a rather different
approach to gathering the vocabulary, namely associations directly from images.
My other use of associative tests is not used by Ørnager, either, this use
being the opposite association from words to images. I experiment with this in
constructing a new tool for designing thesauri. Furthermore, I distance myself
from Ørnager by a more thorough theoretical explanation of the index term in
connection with the Press picture, also to be used in the design of
thesauruses. The thesis contains a large list of reference publications, including the following on-line sources: Arts & Architecture Thesaurus. Online version. URL: http://www.getty.edu/research/tools/vocabulary/aat/.
Neumueller, Moritz (2001): Hypertext semiotics in the commercialized internet. URL: http://sammelpunkt.philo.at:8080/archive/00000023/01/HTML_Version/text/. Citeret/accessed 27-04-2002. Steemson, Michael (2000): The Caldeson Picture Concept Indexing Models. URL: http://www.caldeson.com/moodtop.html. Citeret/accessed 01-05-2002. Stonehill, Brian (1998): The on-line visual literacy project. URL: http://www.pomona.edu/Academics/courserelated/classprojects/Visual-lit/intro/intro.html.
Citeret/accessed 01-05-2002. The original thesis is presented in Acrobat Adobe 4.0 .PDF format. It comprises a file of a little over one megabyte. Downloading can take up to six minutes. Michael Ewan Andreasen, 27. Address: SIgrid Undsets Vej 172B, 9220 Aalborg Ø. Email: wiredloose@myrealbox.com. He successfully attained a masters degree (Cand.Scient.Bibl.) with this Index for Press Photos thesis. He writes: "During my high school years I spent a year in Australia. Lately I have been spending time in Romania. I am now married to a Romanian and know a little Romanian culture and language. These trips abroad, meeting different cultures, has taught me always to view problems from several angles, and I have become aware that things aren’t always what they seem. "During a trainee period and a job in a public library I have proved that I can maintain a satisfactory customer contact to varied types of patrons. I have during my education participated in student jobs that gave me experience with database construction and maintenance (MS Access). I have received good credit for my independent approach to my work in these instances. "Currently, my hobby project is to increase my knowledge of the operating system Linux and transfer my skills from Windows to this platform, including administration, web publishing and database construction. Furthermore, I am very interested in music and an enthusiastic guitar player in my spare time." To go to the RIMOS home page To go to The Caldeson Consultancy main index page 2.5 Thesaurus construction
2.6 Literature and previous research
Go to Chapter 3: Visual Grammar
Internet Links
To open Indeksord for Pressefoto: Designprincipper for en tesaurus
The Author
