4/9/00 abstract
Gombrich, E. H. (1992). The visual image. Scientific American, 221,
86-96.
An image is any graphical representation of something. Images are not equivalent to textual statements. They can contain more information or less information, depending on the type of statement. As images become less realistic and more symbolic, they can become easier to interpret, but once they become more abstract than symbolic, they are difficult to interpret without help. The context, caption, and code of an image are important for understanding its meaning.
4/9/00 discussion
Layne, S. S. (1994). Some issues in the indexing of images. JASIS,
45(8), 583-588.
Krause, M. G. (1988). Intellectual problems of indexing picture collections. Audiovisual Librarian, 14, 73-81.
These articles claim to be about images, but both of them present arguments that could apply to any indexing problem.
For example, Layne divides the possible attributes of an image into four major facets:
If we simply replace the word ``image'' with ``document'', these attributes all relate to general indexing. The articles use images for their examples, but they could have used books, or even antelope.
If I had written these articles, I would have mentioned this:
One major difference between textual documents and other types of documents is that textual documents may be indexed solely based on the words that they contain, while non-textual materials need interpretation for even the most shallow types of indexing. The result of this is that it is possible for a computer to create a (shallow) index for textual documents automatically, but non-textual materials must always be indexed by a human.
4/9/00 discussion
Berinstein, P. (1999). The Big Picture - Do You See What I See? Image
Indexing Principles for the Rest of Us. Online 23(2), 85-86, 88.
Berinstein's point about ``pictures don't describe themselves'' is absolutely correct. That's what I just said above.
It seems that these writers want image indexing to be more thorough than indexing of textual documents. Many books can be indexed by who, what, when, where, generally and specifically, of and about. But we generally only pick a few of the most salient features when indexing a book. Would we index War and Peace with the term ``Russian ballgowns''? Should we?
Do people use images in more ways than they use books?
Berinstein's 3 points to remember when indexing are useful:
4/9/00 discussion
Austin, D. L. (1994). An image is not an object: but it can help. In
A. H. Helal and J. W. Weiss (Eds.), Resource sharing: new technologies
as a must for universal availability of information,
(p. 277-294). Essen: Universitätsbibliothek Essen.
This article discusses the relationship between objects in museums and images in museums and libraries. It also gives some examples of fields used to index images at a few institutions. Otherwise, there wasn't a clear focus, and nothing really memorable.
4/11/00 summary
McCloud, S. (1993). Understanding Comics. Northampton, MA: Kitchen
Sink Press.
(This is more a table of contents than a summary, since the book doesn't include a table of contents, and I'm always forgetting what is in each chapter.)
4/11/00 discussion
Beebe, C., and Jacob, E. K. (1998). Graphic language documents:
structures and functions. In W. Mostafa (Ed.), Advances in knowledge
organization, vol. 6. Frankfurt/Main: Indeks Verlag.
Er, well, this article uses the word ``Gestalt''. I understand the meaning of Gestalt, but for some reason, any article that uses the word is totally incomprehensible. This article was no exception.
4/11/00 abstract
Grund, A. (1993). ICONCLASS: On subject analysis of iconographic
representations of works of art. Knowledge organization, 20, 20-29.
Describes the ICONCLASS classification system for occidental art. ICONCLASS is a faceted system that uses a combination of letters and digits to represent concepts. The focus is on historical and artistic concepts.
4/11/00 discussion
Lunin, L. F. (1994). Analyzing art objects for an image database. In
R. Fidel et al. (Eds.), Challenges in indexing electronic text and
images (p. 57-72). Medford, NJ: Learned Information.
This article brings up some important points. When the image being indexed is artistic, there may be much more to the image than is immediately visible. The image may be indexed based on the reaction it provokes in people, the physical qualities of the object pictured, and many other features. Art objects tend to carry a lot more cultural context than photographs or other images. In general, images are more difficult to index than textual materials because they don't explicitly tell us thier purpose (lacking Gombrich's context, caption, and code).
4/11/00 abstract
O'Connor, B. C. (1991). Selecting key frames of moving image
documents: a digital environment for analysis and
navigation. Microcomputers for Information Management, 8, 119-133.
ISIS analyzes a moving image document and constructs several types of ``abstracts'' for it. The document is broken into segments based on the percentage of the image that changes between successive frames. The lengths and contents of segments can be compiled into a type of abstract depicting the spatial-temporal content of the document. The frames on either side of a point of discontinuity are selected as key frames, which can be combined into a pictorial abstract.
4/11/00 assignment
Working within the representational parameters provided by Layne
and/or Krause, create access points [index terms] for each of the
images. Record the list of terms n your journal and indicate whether
each of your index terms reflects biographical aspects of the image,
its objective content (the "of-ness"), its intellectual content (the
"meaningfulness" or "aboutness") and/or the potential usefulness of
the image. What were the difficulties you faced in generating these
index terms.
(B)iographical
(O)f-ness
(A)boutness
(U)sefulness
Image 1 - black (O), abstract (U), 1993 (B), Smithsonian (B), face (O), smooth (U), head (O)
Image 2 - 1993 (B), Smithsonian (B), gold (O), silver (O), jewel (O), wealth (A), religion (A), elaborate (U)
Image 3 - 1993 (B), Smithsonian (B), travel (A), parking lot (O), cars (O), transportation (A)
Image 4 - 1993 (B), Smithsonian (B), pastel (U), washline(O), outhouse (O), wind (O), west (O), homestead (A), loneliness (A)
Image 5 - church (O), worship (A), 1993 (B), Smithsonian (B), simple (U), black (O), family (O), cart (O), ride (A)
Image 6 - trees (O), 1993 (B), Smithsonian (B), building (O), garden (O), people (O), fountain (O)
Image 7 - green (O), pattern (O), woven (O), 1993 (B), Smithsonian (B)
Difficulties
4/17/00 group work
We looked at the images again, this time with the museums'
descriptions at hand. I'm not sure what we were supposed to get out of
this, but I was actually surprised at how well I had interpreted the
images without the descriptions.
Jake and I spent some time looking at other image search services and playing around with thier systems. Jake sent a question to Corbis on their indexing procedures; I don't know if he received an answer.
4/17/00 class notes
A title constrains how an image is interpreted.
Jackson Pollack used numbers to title his abstract paintings, so ``the one with the pink and grey dots'' could also be ``the one about infinity'', or whatever the viewer happened to see.
We may be able to accurately describe the ``of-ness'' of an image, but the ``aboutness'' depends of the viewer. In fact, the whole concept of ``aboutness'' is up for interpretation.
Abstract indexing terms assume a meaning for an image that may not actually be there.
In general: