What’s the best way to leverage the most information out of 140 characters? Should you get to learning Mandarin so each character can be a word? Or start forming German-style pseudo-word hashtags to get the point across? Or perhaps, you could parse the natural language, encapsulate the tweet in meta data and go from there.
We’ve already seen additional information stacked onto our Tweets, as with the geo-location API released last November, but Cascaad’s SuperTweet API does more than wrap your tweet in client-provided data like GPS coordinates.
Cascaad has released its first beta of the SuperTweet API, which it says will allow third-party Twitter applications to “add smart contextual information and monetization […], including semantic entity markup, nonintrusive in-text affiliate commerce links, related content [and] social relevance scores”. The SuperTweet provides users with “an at-a-glance view of additional information about stories, things and places discussed in the message, without forcing them to leave your application,” according to the API documentation.
The API allows developers to parse a tweet, identify separate “entities” and then gather external contextual information on those entities. It then adds this information to the original tweet to create a “SuperTweet”.
If a tweet mentions Lady Gaga, for example, the name “Lady Gaga” becomes a link to a biography pulled from semantic-web database Freebase. Next to that, the SuperTweet gives an affiliate link to Amazon, where you can go buy Lady Gaga CDs. And if a link to an article about Lady Gaga is included in the tweet, the SuperTweet provides a thumbnail preview.
In addition to wrapping these entities in contextual information, the SuperTweet API unwraps shortened URLs back into the original link so the user has an idea of what they’re clicking on. And, although not yet available in this release, the Conversation API will put the tweet in the context of a conversation, providing access to other public messages in the same conversation thread.
The challenging part of all of this is that the API needs to parse a rather variable piece of content – a user created tweet – and find the appropriate meta data. Just like a search engine, it needs to recognize misspelled words or other slight variations to find the proper content.
One Twitter developer we spoke with said that, while they like the idea of outside information being added to the base tweet, they have found the contextual results to be hit or miss. It would seem that the concept is solid, but the execution is still in the difficult learning stages.
While we like what we’ve seen of the SuperTweet so far, it will only be worthwhile if it can provide accurate results. If we tweet about the iPhone and it links to the Amazon page for the iPad, the service will fall flat on its face. Get this part right, though, and we’re willing to be you’re going to start seeing Super Tweets in some Twitter apps soon.