I enjoyed this blog post on scene-level television metadata, more than I thought I would. I don’t own a TV or watch very much TV these days. But reading this post made me realize that I am perhaps part of this new breed of audience they refer to that gets their TV fix from the internet and has to hit pause every now and then to IMDB something and so on. While the discussion of how tagging TV is a new and profitable trend in the metadata industry and how it’s not as easy as it seems like it should be because of lack of standards, etc. was super interesting, what captivated me most was the context for why all of a sudden there is such a demand. The stats about how many people are connected to the internet or are distracted by their smart phones while they watch TV, along with how traditional TV advertising is no longer making money all makes perfect sense and is stuff that I hadn’t really considered much until now. What worries me though is thinking about what kind of lengths advertisers will now go to push their product and how this will interfere with my entertainment and distraction. I hate product placement in shows and movies and I hate feeling like I’m manipulable in that way. As traditional commercial breaks die out, will ads just become increasingly more subliminal and insidious? Blecchh no thanks!
I feel like I am starting to sound like a broken record about this “Relation” element in that I still don’t know what we’re going to do with it for our indexing project. My online research has yet to reveal any answers that I think would suit our particular scenario. Many of the repositories that are linked from the project wiki don’t use Relation. Also I haven’t come across any helpful sports image databases where I can view the metadata and see if/how they use this element.
On the bright side though, because of all these factors it’s beginning to seem to me like we can use Relation however we see fit. While Dublin Core best practices highlight a list of refinements to help specify what kind of relationship is present (IsPartOf, IsVersionOf, etc.), I’m beginning to wonder if we couldn’t just create our own all encompassing qualified standard that states that such and such image “IsRelatedTo” followed by the unique identifier and link to whichever image(s) we want to relate. I know that this is a little vague, but since all of these items are images coming from the same place, can’t the specific relationship between 2 or more photographs that are taken within milliseconds of each other (as in images 3 and 4 on this example page) be inferred? I also like this solution because it’s easy and straightforward and doesn’t require the more football challenged among us to invent any new terminology and therefore I think is less susceptible to error.
Before I go any further with this though, I’m also starting to wonder if it will become more clear when we’re finally able to see the images that we are working with. But for now, this is my idea. What are your thoughts?
So I did my best to follow along with the hour-long LD4L presentation delivered for the Coalition for Networked Information last December that Dr. MacCall emphasized to us at the end of Wednesday’s class. Overall I think it helped me get a better grip on some of the advancements and challenges of linked data collaborations between libraries, though I have to admit that a lot of this is Greek to me. LD4L (Linked Data for Libraries) is a collaboration between Harvard, Cornell, and Stanford and I think what’s most impressive for me is the staggering amount of resources they are trying to capture and account for through this project – 13 million for Harvard and 8 million a piece for the other two. Some of the examples they gave for how this linked open data initiative will enhance relevance ranking, research, and discovery really resonated with me. In one they discussed two faculty from different universities who have a shared relationship in that they are both affiliated with the same outside organization. Through their linked data vision though, users will not just be able to see this mutual affiliation but also will be referred to each other’s articles when searching for articles by one or the other. Now imagine how quickly this could be expanded on. One of the biggest challenges that stuck with me is this idea that no one is right, no one is wrong, and it’s not quite clear what the objectives exactly are all of the time. I liked how they summed up the problem of entity reconciliation with the question when do we mean “Same As” vs. “See Also”? (i.e Stephen King and Richard Bachmann or the example they used Mark Twain and Samuel Clemens.) The answer is that a lot of the time we still don’t really know. After watching the video I cannot now say that I am an expert in this stuff, but it’s a really interesting topic and it seems like there are some pretty brilliant minds coming together to work these issues out.
During the breakout session in class tonight our image indexing group talked briefly about what we are all thinking for drafting our guidelines. Before spring break as a class we decided that my element, Relation, would best serve the collection by denoting images that are related sequentially and that are derived from the same play. No other element will directly do this, and it could be a useful piece of metadata for search. How to know that this relationship exists between any 2 or more images is the rub.
Since we are indexing in only “semi-automatic” mode, the project manager (in this case, me) will notify everyone in class when the assigned images are sent out whether or not any of their images are related to each other. Therefore, no one will really need to make this judgment call on their own. If we had the capability to be fully automatic, however, Amy pointed out that this kind of relationship could be made known via the image time stamp. Definitely a handy idea.
Still it appears for me at least that the hard part is not over, since I am the one who has to come up with the indexing guidelines for this element. I’m having a tough time thinking of an easy and standard way to phrase what the relationship is that’s taking place. “Image 1 is part of the same play as Image 2″? Is simply “Image 1 is related to Image 2″ sufficient? Nikki had the brilliant suggestion of looking at game transcripts or play by plays for terminology, and I wonder if there are any sports image repositories out there that use the Relation element in this way. I definitely have a little work cut out for me. If anybody else from the class wants to weigh in or is seeing something that I’m not, I am all ears!
Alright, back to the old WordPress…
Last week was spring break and it was a good thing that I was able to get some rest because I spent the last four days in Fort Worth, TX at the Art Libraries Society of North America (ARLIS/NA) annual conference. I was totally honored and completely psyched to be this year’s recipient of their internship award, and thankfully I was able to put together some travel funds from UA and SLIS in order to attend. This being my first library-specific conference (though not my first professional or academic conference), I was somewhat nervous of what to expect. At the risk of hyperbole though, I found it to be a truly invaluable experience. Everyone that I met was super encouraging and supportive, and I can tell you now that I feel so validated, inspired, excited, and optimistic about my upcoming internship and my future in this field. I sincerely hope that everyone out there has had or is able to have this kind of experience at a conference in their field soon!
Obviously I couldn’t go to everything, but I checked out a lot of really great panels. Topics covered asset management, web archiving, collecting artist books/photobooks and artist “recordworks”, controlled vocabularies in foreign languages, and documenting art through social media. Many of the points discussed and questions asked hit on some of our favorite metadata issues. Related highlights for me included:
1. A “New Voices” panel which had a PhD candidate from UW Madison discussing contemporary Chinese artist and activist Ai Weiwei’s use of Instagram and Twitter to document his artwork. Librarians need to come up with a method for making this primary source material accessible and appropriate for student research.
2. A librarian from the Universidade de São Paulo discussing her collaboration with the main SP art museum to update and improve a controlled vocab in Portuguese. I have an MA in Latin American Studies with a focus on Brazilian art and literature so this one was right up my alley, and it also addressed pertinent issues for us in our class like how to translate concepts for which there aren’t any good equivalents in other languages.
3. A Q&A following a panel on photobook collecting that raised the age-old problem of whether to catalog as “artist book”, “photobook”, or both, and one librarian’s revolutionary effort to let the artists themselves choose their own metadata, if you will, through a brief form he gives them at point of sale.
Lastly, I had the excellent fortune of getting to attend the conference when the keynote speaker was the director of the International Center for the Arts of the Americas, the Latin American art research wing of the Museum of Fine Arts Houston, and where I am hoping to complete my internship in Fall 2015. It was a great talk! What a blast! And now I’m ready for bed at 3:30 in the afternoon. Thanks for reading!
Leave it to the next article I read only after I blog about something to give me the term that succinctly describes one of the observations I was clumsily getting at in the post. Beneath the Metadata introduced me to the term “meta noise,” which is used to refer to all the random, irrelevant, misspelled, or otherwise weird and quirky tags that I reflected on (without having the term for it) in my previous blog entry. The unsystematic lack of control that leads to meta noise and tags that contradict each other is what makes folksonomies problematic in Peterson’s estimation. However, 9 years after the publication of this article it seems from my own observations like the type of tagging that would qualify as meta noise has become more the norm. I think the persistence and popularity of these tags prove their merit. Could it be time then to stop dismissing it as “noise” and start more seriously considering the weight and impact of these tags?
Also, this felt relevant:
Joan Beaudoin’s “Flickr Image Tagging: Patterns Made Visible” got me thinking about some trends that I’ve noticed in social media image tagging. I’ll go ahead and make an honest mention here that I personally don’t participate much in this practice nor do I regularly post images or my own photos on social media, so maybe my observations don’t hold much weight, but as always I am curious to hear what others think!
This importance of contextual knowledge that Beaudoin refers to is to me the most critical aspect when it comes to tagging. More and more it feels like whenever I see a hashtag in one of my feeds it’s followed by an inside joke or words strung together in a clever way that don’t really mean much without the image association (and sometimes even then the meaning is lost on me). While often these are topical phrases or slogans that one can use to easily contextualize (e.g. #RollTide, etc.), much of the time these kinds of tags strike me as one-offs that only serve to make a joke and I don’t believe are meant to be a future searchable category. I suppose this is a reason why some information specialists dislike tagging? (Or maybe I need to spend more time on social media?)
On the other hand, another trend I’ve seen that I find easier to get behind is how users tag certain images with their own word, quote, or phrase that then becomes like a personal handle for the different kinds of images they post. I feel like this practice might be one unexpected consequence of an infrastructure that supports social tagging. Not only can you search by user, you can also search by that user’s unique hashtag. This trend and the one mentioned above lead me to wonder what effect this type of tag-coding will have in the long run. It’s like you’ll need to be fluent in a language that’s specific to a small group of people or even just one person in order to retrieve anything.