Posted by: Emily | April 8, 2010

Semantic web what?

I spent my morning at a METRO-sponsored talk by NYU’s Corey Harper about the semantic web. I’m a smart cookie with strong reading comprehension skills, but there’s something about this semantic web thing that I just can’t understand. (I also can’t draw maps; I suspect the problem is a verbal-to-visual translation error in my brain.) I’m not a cataloger or a metadata specialist and probably never will be. But I am interested in the ways classification structures and controlled languages are productive and re-productive of dominant knowledge regimes. Usually I think and talk about LCSH in this regard, but if the semantic web is where we’ll all be looking for and finding things in the future, I wonder if I should shift my focus.

So, some provisional takeaways for me from this morning’s session, with all due disclaimers about how much I still have to learn:

  • The semantic web has lots of fancy words and is way more elaborate than the records I’m used to seeing, but still consists in large part of developing classes and identifying ‘things’ that are instantiations of said classes and then naming them. So the critiques we have of systems that mark some things (woman, of color, poor, immigrant, etc.) while leaving other things unmarked (man, white, rich, native, etc.) should still hold in in this brand! new! way of organizing data on the web. Or not?
  • The semantic web is concerned with relationships between things. In other words, it’s explicitly concerned with relational structures beyond the BT/NT/syndetic structure of LC. That’s a lot of standardization of things that are less fixed even than objects, right? I’m still puzzling about what the implications of this might be. I’m thinking about relationships that make sense at one point in a given knowledge formation that we would not want to make persistent. Would the semantic web of 1911 make permanent the relationship between Africa and bestiality? Presuming we’re no better as humans today, what kinds of  ‘common-sense’ relationships will get codified that might be better left un-named?
  • You can draw graphs, or representations of all the connections between pieces of data. I had a professor tell me once that if you couldn’t visualize it, it couldn’t be. I also think of the number of things I understand better after reading Foucault’s formulation in language, counter-memory, practice that figures limits as waves on the shore, drawing that shifting line in the sand. So, maybe the graphs offer an opportunity for us to begin to see in new ways the underlying structures of our data, giving us new texts to read for analyses of power.
  • I’m always worried that with privatized algorithms and the general opacity of computer science that we won’t be able to “read” the structures that retrieve some things and not others when we do our searches on the web. How will we understand what we retrieve and what we don’t and the reasons why if we can’t see the bones of the information structure? I liked that a lot about the semantic web examples Harper showed us–we could see it. We looked at the metadata for web-based representations of Bob Dylan. I could see right away what the producers of that web information marked (musician, member of various bands, etc.), what they didn’t (Dylan’s gender, race, etc.), and what they marked in ways that didn’t attend to culturally different significations (what folk music means in one place is not what it means in another). So if my hunch is right that we might read classifications for a better understanding of the politics of what counts as knowledge, the semantic web certainly seems to be productive of texts we can all start (and continue) reading.

Last aside before my desk shift ends: I also went to a talk last night by Dean Spade about the limits of legal reform. (Smarty.) The law’s another place where categories matter, and his analysis was provocative and we can talk more if you want to. In this minute, though, I wonder what kinds of interesting conclusions we could draw if we brought Spade’s analysis of Foucault’s conceptions of governmentality and biopower to an analysis of the knowledge regimes on the semantic web and, of course, in our libraries.

Advertisements

Responses

  1. Thought-provoking post. Thanks for shaking my brain!

  2. Thanks for this post, Emily. You’re comments about permanence of relationships were something I hadn’t thought about.

    I came away from yesterday’s conversation seeing semantic web* as progressively disruptive of hierarchies in classification and the relationships those hierarchies preserve. As we heard, much of what is being done to link open data is being done from the ground-up, not by dominant institutions that reinforce existing power structures. That opens up issues of trustworthiness (for some), but I’m inclined to err on the side of having more voices involved in that process — including, but not limited to libraries.

    *I also think we might do well to get away from “semantic web” as a term. It’s got way too much baggage for its own good right now, and “linked open data” seems much more transparent and accurate.

  3. Thanks for the engagement, Jason.

    Linked open data = yes, yes. Much clearer, and I feel like it says what it says it is, which as a layman I appreciate.

    And yes, I agree that the linked open data movement is definitely more multivocal and flatter and more amenable to posing alternative meta-narratives than the top down library classification hierarchies. And yet, when we looked at the ‘bubbles’ of open data, where did they come from? U.S. Census. BBC. NYTimes. Very much from concentrations of, at the least, technological capital.

    Also, I think whoever ends up doing the naming and the linking and the collating, even when its done self-consciously and in recognition of the power of metadata to create dominant narratives and knowledge formations (like at the Prelinger library, I think), it still ends up telling a limited range of stories that leaves out some and includes others. It’s just a material constraint on the project of organizing information. So I guess what I want to suggest is that we still attend to the nature of those stories rather than celebrating any classification structure or system as good, or equal, or right.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Categories

%d bloggers like this: