To kick off the readings this week I got into some image indexing theory as well as some pretty brain-beating discussions of ofness and aboutness. I specifically enjoyed the theoretical history and overview contained in the “Image Indexing” article by Neugebauer in addition to a couple of entertaining and illuminating blog posts by my classmate MetaDamen here and here. One idea that I really gravitated to from the Neugebauer piece is that image indexing is somewhat akin to accommodating Goldilocks (forgive me for this comparison here) in which the objective is to present images at neither the most advanced nor most basic level of constructed meaning but rather at some intermediate level in between (to paraphrase Jorgenson). J. P. Small went so far as to coin this principle the “preservation of ambiguity”.
This concept was at the time unknowingly put into practice by me last semester in Dr. Albertson’s Digital Libraries course where we were required to create a mock digital library from a set of old Alabama football images. To be honest I’m not totally thrilled about having to work with these images again for this class. I follow football a little bit but not too diligently, and most of all being a non-native to this region I shamefully know almost nothing about Alabama football (really, beyond Bear Bryant and “Roll Tide” I can’t offer you much). As a result, when inputting my Dublin Core metadata terms for this project there was only so much of a description for what was going on that I could give. An example that I can remember clearly is a photo of some coaches conferring on the sideline during a game. I could deduce that what I was looking at here were indeed “coaches” and not merely “men in polo shirts with headsets on,” but I could not tell you specifically what these coaches’ names were nor what specifically was happening at that moment in the game.
I feel better knowing that there is a principle acknowledging this type of behavior. It always worries me to think about bias that indexers could inject because of their personal knowledge of or experience with a certain subject. I think what is super important and still somewhat of a mystery to me is how to utilize and adhere to a controlled vocabulary and to do it in a way so that users are on the same page. I’m looking forward to learning more about tagging so watch for a post about that soon!
Adam's LS 566 Metadata Course Blog 
I, too, am glad that “preservation of ambiguity” is a thing. When we did that project for LS569, I had my uncle and cousin help me with image description. I remember one specific picture that was of the quarterback. I asked them how they knew he was the quarterback, and the answer was something about knowing football (which I don’t remember exactly, but apparently number ranges mean something?). I think that’s part of why we need rules for indexing within a collection. My images would all say “football player doing something”, and his would say “quarterback calling out play” or something equally more useful.
LikeLike
This is very intriguing. I find myself dealing with this dilemma more often than I realize now that I consider it, both personally and professionally. Good post.
LikeLike
I struggled with this as well. It was only after hours of Googling and careful checking of all the old team rosters (aha! that guy is the “running back” … I think that’s important!) that I could get beyond your basic “Alabama player running away” or “player number 61 does chicken dance while holding football.” In the end though, maybe it fit that ambiguity principle rather well–I put names to numbers and identified actions like throwing, catching, and tackling. There wasn’t really enough context for anyone to definitively say “QB throws for a touchdown” or anything else that specific, and you could argue that maybe that would be more specific than most people would search for anyway.
LikeLike