FOAF to the People - position paper for FOAF Galway
Maybe it's the smiley logo. Maybe it's
"dnaChecksum." Whatever the reason, FOAF and friends
are a breath of fresh air. The business and research worlds
are filled with nitty-gritty aggrandizing, "look at me"
positioning, and plenty of people taking themselves too damned
seriously. FOAF offers something lighthearted and immediately
useful, with power to spread like wildfire. FOAF also embodies the
best qualities of the Semantic Web: loose, light,
interconnected, open. We see FOAF as setting the tone for
other efforts, helping kickstart the task of motivating people to
annotate and interconnect everything. Before that "tipping
point" happens, we gotta get regular, non-technical, people using
it. That's our focus.
2. Tidepool & Storymill
Immuexa is developing two products that use FOAF. The
first is a desktop application called Tidepool™ that allows
people to annotate photos and other media, then browse and combine
them into stories. Annotated media can then be shared on our
second product, Storymill™, a website application and service
that serves as a showcase, discussion forum, and central hub,
allowing Tidepool users to replicate their memories around, using
RDF for metadata. Both products also use instant messaging and
"precognitive" agents as central to their architecture and user
3. Storymill Accounts
To sign up for Storymill, visitors are asked the usual questions to create their account, after which a FOAF file and PGP keypair are created automatically. You can sidestep these questions by supplying an existing FOAF file, but you cannot supply your existing PGP key, as our use of PGP is currently Storymill-specific since we want that private key for signing and encrypting media for use in the web-of-trust.
Our goal is to get lots of people making both FOAF files and PGP
keypairs without them even knowing they're doing it. Our use
of FOAF and PGP is completely transparent to the user, unless they
know to look. Both the FOAF file and public PGP key will be
available on Storymill in a predictable location, making things
easy for FOAF trawlers and the like, though if privacy is a
concern, you can opt for an "unlisted number."
4. The Mu Schema
Besides account information, we're using FOAF as a well-integrated companion to our "mu" (myoo) schema. Why call it mu? Well, we've found about a dozen meanings for mu, such as "don't ask" in Chinese. The silliest we've heard is that it's a new pronoun for "me and you." Somewhere in the middle lies the mystic nature of mu.
The mu schema should be thought of as an adapter (or bridge) schema, not a replacement for other efforts. Our intent is to describe the simplest architecture possible for describing memories and media, then use it as a connecting point to other schemas. We want to interop with everybody.
The core classes are mu:Location, mu:Item, mu:Link, and mu:Tag (LILT). Items are media resources, such as photos and text blurbs. Items correspond to FOAF Documents. Locations are places you can find digital media, such as Storymill and other websites. Links are explicit relationships between Items, Tags, and Locations. Tags are annotation nuggets that can be attached to Items. Though these concepts certainly overlap other ideas (properties, URLs, hyperlinks), we've got good reason for making them describable resources in their own right.
Crucial to the architecture are the four tag types: who, what where, when. We break everything you can say about a media item into those four categories. This "4w" approach has many implications, especially increased usability and comprehension for non-technical users.
Tag types also play a big role in how we integrate other schemas
into mu. "Who" becomes a wrapper for foaf:Agent, vcard, and
other PIM formats. "What" will connect to Wordnet and other
categorization efforts. "Where" fits nicely with the geo/map
crowd. "When" connects with calendaring.
5. URI Syntax
Also of interest is our URI syntax, which is the least FOAFy part of our plan, as it strictly references our resources, unlike the mbox approach. Resources created with Tidepool or Storymill get a URI of the form:
where "class" can be ( location | item | link | tag ) and "seqnum" is a number relative to that class for that user on that host. For example, a photo called "http://storymill.net/mu/timothy/item/123.jpg" would be the 123rd item for user "timothy" on "storymill.net".
Replication is a big part of Tidepool and Storymill. Media and
metadata are easily "gathered" and "shared" between Tidepool users
through Storymill, allowing collaboration via the web and through
Tidepool's IM chat interface. Media and metadata will also be
accessible from Storymill via HTTP using the same URI syntax,
allowing other systems to easily connect with us.
All very good, but what's the fun in URIs and replication? Not much, unless you're in the angle-bracket crowd. Given that our primary goal is to motivate the masses into annotating everything, how do we spark their imagination and woo them to do the otherwise laborious task of tagging?
With a game of course! "PhotoBingo" is a web game that lets players fill out 4x4 randomized bingo cards with annotated attributes as chips and "who, what, where, when" at the top instead of B-I-N-G-O. We're premiering it at the Galway workshop, giving tee-shirts away as prizes.
The game website will continuously aggregate a variety of annotated photo feeds, including those from Storymill and w3photo. As attributes show up in the feed, the website will "call them out" as a bingo caller would: "Who-danbri", "what-sparrow", "when-Tuesday", "where-Ireland." When players complete a row, column, or diagonal, they'll click the big "Bingo" button to achieve dubious glory and a spot on the scoreboard.
Yes, it's a dorky game, but very much in the spirit of FOAF, we feel. We're keen on anything with potential to cause PhDs to roam the streets looking to photograph "what-bicycle" or "where-QuaysBar." It could be a lotta fun.
More details can be found at http://photobingo.com.