Bitwacker Associates

Earlier today via his Twitter feed Nova Spivack started to brain-stream about a Semantic Twitter Client (Semtweet) In a feed about two hours prior to me starting this post, Nova jotted down some thoughts:

Semantic twitter client vision: like firefox for twitter, extnsble, detect topics & types, linking, threading, filtering, metadata

I think this is a great idea! In today’s post I’d like to riff on Nova’s “spec” a bit…

• Like Firefox for Twitter…Above all else, to me this means it would be supported by an open source community. From this derives other attributes, including: multi-platform; extensibility (see also below); and shipped as standard on most platforms. I would add one critical UI feature that most of us first saw with Firefox: tabs! Tweetdeck’s columns are useful, but there are times that I wish I had tabs…
• Extensible… Nova’s Semtweet will achieve maximum value if it is conceived and executed as a platform that attracts a community of add-in developers. And to achieve the greatest mindshare, extensions should be easy to implement, ie they should be more like scripted GreaseMonkey-based extensions or iGoogle gadgets than “full” Firefox extensions.
• Detect topics and types… This I suspect is the “killer” feature that Nova wants to “major” on, and the way that the underlying functionality is integrated with a novel UI metaphor could make or break the Semtweet. Also, it is the one feature that makes me wonder, are we really talking about a client or should the Semtweet actually be a service? If the latter, at least in prototype form, then advanced UI code such as much of the SIMILE toolkit could be mixed and matched with topic detection/extraction to create some insanely great real-time renderings.
• Linking… Tweets are more useful for both readers and authors when links have been embedded. A next- generation, Semtweet should not only make embedding links easier, but should automatically sense and aggregate related links to augment individual tweets and threads and should “surface” these in an advanced UI. A more advanced feature would facilitate the embedding of references to hosted aggregations of links, using a named graph and/or OAI-ORE based approach.
• Threading… One of the insanely great features of a client like Tweetdeck is the rendering of threads as columns based on arbitrary keyword-based searches. It works so well that it is hard to remember that all of this is virtual; in reality, every tweet is thrown into the soup with all others, but clever tagging and viral “best practices” help us make sense of it.
• Filtering… I can see a need for both positive and negative filters. Positive filters would work much the way keyword searches in e.g. Tweetdeck do now; negative filters would work more like junk/spam filters in Thunderbird. Hmmm, should Nova had specified, Like Firefox and Thunderbird for Twitter?
• Metadata! This is my favourite of the facets Nova listed. It’s not clear to me whether Nova is refering to deducing or otherwise extracting metadata from tweets and threads; embedding metadata (somehow, like hashes) in tweets; or facilitating the association of halos of metadata with tweets. Frankly, this last option is most interesting to me! Since each tweet as a unique URI (for example, see http://twitter.com/novaspivack/status/7442269445) a useful service would allow the associated of arbitrary metadata with a tweet’s URI, somewhat like delicious. By itself such a service would be painful, but seamlessly integrated with a Semtweet could be very cool!

I’m being purposefully brief today, to get these ideas “out there.” Please comment, tweet, and link! 😉

Updates: Since this is an active thread, I will post updates as appropriate:

• As I was writing this, Ed Borasky made note of The Twitter Streaming API and Why It’s Important
• As per @novaspivack, use the #Semtweet tag to follow the Semtweet Twitter discussion
• In August 2009 TechCrunch wrote about a tiny thread, created by delicious.com founder and current Google employee Joshua Schachter.
• On 7 Jan 2009 the discussion turned toward semantic microblogging, with the distributed microblogging prototypes SMOB (Alex Passant, et.al. DERI) and TwitLogic (Josh Schinavier, TWC, RPI) being highlighted amongst others. More on this in an upcoming post…