Tag: metadata

Posted on

“That Movie with the Sliding Doors and Time Travel … You Know…Maybe in Polish”

Book store staff, librarians, and movie store clerks (back when there still were movie stores!) all have stories of perplexing requests from patrons, generally when there is a very specific title in mind, but the question is a complete puzzle, with a title described in such vague or idiosyncratic ways as to require mind-reading more than reference desk skills.

A few tidbits from Library Thing,

” I need that book that’s called Shakespeare, but it’s spelled with a “Ch” and the author starts with M…”

Fortunately, I was in my groove that day, and it only took me a few seconds to figure out that the patron wanted the book Chesapeake, written by James Michener.

••••••••

My friend, as a young lass, once ran up to a librarian, very excited, and yelled out, “Do you have ‘The Cat Who Shat?'”

••••••••

I worked in a record store all through my (many) undergraduate years and we would get many crazy folks into the store on a regular basis. The overall most common silly request was; ” Can you help me find this song? I don’t know what it’s called or who sings it, but it’s about love.”

Aren’t they all?

As a classical music person, I’ve encountered my share of these mysteries: Somebody raving about hearing “Faust’s Requiem.” (I figured it was some outrageous East German world premiere, but it was just the Fauré one.).

When I was responsible for an information service on opera (which is a fairly nutty thing to start with), I did reliably got calls about “that opera where she dies at the end.”  I checked my impulse to say, “if she doesn’t die at the end, it’s  probably just Rodgers and Hammerstein.”

hunt Library Stacks.sm
The automated, as in robot served, stacks at NC State’s Hunt Library. A lot of metadata within and without. Maybe someday we’ll be able to tell a robot, “I want that blue book, about a whale named Ahab…”

But on reflection, all these misfires disclose that the brain is an interesting thing, and though no doubt we may all share much common neural architecture, we certainly don’t seem to keep track of and remember things or their associations in the same way. (In my case, I would charitably call this “neural metadata diversity of recall” but I suppose it could just be that I’m ‘getting on’ as the Brits would have it.) There is after all, a specific target for all these descriptions, it’s just that the getting there is fraught.

For myself, I do okay remembering  music facts (I started early in this domain, and, since I listen to and play music every day and write about it almost as much, it remains relatively fresh), but boy do I get foggy about movie details. And am notorious for groping to remember titles of films I have seen or want to see. “The one about the farmer who drives to someplace in the upper Midwest to collect his life insurance pay out starring the guy who was in The Great Gatsby.”

“Train doors close and open on two different paths for the story.”

“Brad Pitt turns into a baby. Unbearably long “

Turns out, there’s an AI-powered web site that makes this a snap. “What Is My Movie?” assesses deep content using technology from cognitive science to suss out those elusive links in your query and unravel the mystery of which movie you were looking for on Netflix.

From their website:

Whatismymovie.com is a descriptive movie search engine that was originally created as a showcase for Arctic15, Helsinki in 5/2015. Its purpose is to present a new way of searching video content, using movies and TV as the chosen approach. Descriptive movie search is based on our research on what is called “Deep Content”. Deep Content is everything we can see and hear in a video, but cannot mechanically analyze – until now. Deep Content includes transcripts, audio, visual patterns and basically any form of data feed that describes the video content itself. After analyzing the deeper levels of the video, we automatically convert it into advanced metadata. This metadata is then processed by the beating heart of our engine: a cognitive machine learning system that understands natural language queries and matches it with our metadata.

It’s not foolproof. It got Sliding Doors and the (completely intolerable) Benjamin Button in the examples above, but didn’t realize I was thinking about Nebraska for the first. (Although it did suggest Borat!).

Posted on

The Age of Metadata

As the NSA stories reveal, it really is all about the metadata now (if you updated “Our Bodies, Our Selves” for 2014 it would have to be “Our Metadata, Ourselves” or more accurately, “Our Selves, Google’s Metadata”).

One specific species of metadata that has always interested me–a library geek from day one–was citation, and there is an interesting buzz of activity around making citations, particularly in science domains, open. This is a tentacle of a long-running battle in scholarly publishing about the enormous power of big publishers vs. individual scholars/scientists and the latter’s desire to make content open and available. It may come as a surprise that not only the journals and articles themselves are often tightly controlled, but the metadata–who is cited, how often, and by which other authors–is also typically proprietary.  This means that doing broad analysis of this metadata, for instance mapping scholarly knowledge as reflected in the written record, is pretty hard. One example is measuring impact or influence–oft-cited papers (even if is to dismiss the arguments contained therein) raise your academic rank. But that, like the lit review, which these tools also support, is only the beginning of what this “meta” level of citation can yield.

A recent issue of Nature had news of a couple of interesting tidbits about this: first the Open Citations Corpus, which is trying to establish a framework for accessing citations from publishers and other sources. (Now it’s focused on biomedical content, but they have ambitions to scale to encompass every discipline, including arts and humanities.) They are already providing the tools to do this, and Nature is following suit.

A related (and even more interesting) effort is figshare: a way for scientists to publish and share data without the intermediaries of legacy publishing channels.   From their web site:

Who/what/when/where/why is figshare?

figshare allows researchers to publish all of their data in a citable, searchable and sharable manner. All data is persistently stored online under the most liberal Creative Commons license, waiving copyright where possible. This allows scientists to access and share the information from anywhere in the world with minimal friction.

The site offers upload times of mere seconds for all file formats, providing a citable, searchable endpoint for researchers. figshare offers unlimited storage space for data that is made publicly available on the site, and 1GB of free storage space for users looking for a secure, private area to store their research. Users of the site maintain full control over the management of their research whilst benefiting from global access, version control and secure backups in the cloud.

figshare was started by a frustrated Imperial College PhD student as a way to disseminate all research outputs and not just static images through traditional academic publishing. It is now supported by Digital Science, a Macmillan Publishers company.

It’s based on Creative Commons Licensing, and I’m struck that it acknowledges from the get-go that modern scholarship and research should provide a way to capture and publish more than just text and figures. We live in a multimedia world, and use those tools to think and build knowledge. Yet we can’t keep track of the assets that are created with these tools very well.  (If there really is any kind of functional scholarly apparatus for multimedia when used in disciplines other than those about a particular format, I haven’t heard about it.)  When scientists use an asset that is in film format, say, in scholarly publication, how do they catalog, cite, and store? How does it become available in the literature. How do datasets get “published,” for that matter? Is it just, “that data file I sent to Sudhir last month. Where did that go after that, do I still have it? did the data change?”

figshare is one of the efforts to answer these sorts of questions. And it sort of shows that there is a science of data and its management, analysis, and place in knowledge emerging  (“International Journal of Metadata Studies, anybody). Were this library school drop out more ambitious, I’d try to get it going. For now I’ll wish it all well, and blog about it now and then. (Along with the observation that “figshare” is a tad twee for a title, as is the all lower case.)  Like Thomas Pynchon, the Internet does love its funny names.

catalog_card
Metadata 101, from the 1930s, when the Catalog Card was the ne plus ultra of metadata. Found on the Flickr stream of http://ccdl.libraries.claremont.edu/, Claremont Colleges Digital Library, a fun browse.