Twitter And Knowledge Discovery

You can discover four types of Knowledge (from Twitter). The classification is inspired by Bloom’s Taxonomy.

1. Factual Knowledge
2. Conceptual and Meta Knowledge

Let us briefly look at each type with a couple of examples.

Factual Knowledge- Some observations

• You get more news from Twitter than from news papers
• You get more real time news than any other single medium
• You get more reactions to news  faster and reactions to reactions (second and third wave)
• Hot news pops up as trending topics making it more discoverable
• Patterns of information propagation increases your meta knowledge (who, what, why etc.)

Conceptual and Meta Knowledge

In addition to bits of information, you also get higher levels of knowledge, if you just take the time to analyze it.

• You can glean inter-relationships and Structure by analyzing followers, lists and retweets and other referral formats. For example, by looking at the people and lists that thought leaders (like Tim O’Reilly) follow you can get some sense of the information relationships. By looking at the lists Tim is in, you can also understand a lot more of his following.
• You can identify influencers and experts in various topics and industries. Klout tries to do this a bit. You can look at the reach and network effect of certain people on Twitter. You need to augment this analysis by looking beyond Twitter, but Twitter gives you some great starting points.
• Every one on Twitter is reachable (for example, if you want me to notice something, you can just add @dorait in your tweet, drawing my attention. If the information you share makes sense to my audience or emotionally appeals to me, I may retweet it. Guy Kawasaki once mentioned that he looks at Tweets where he is tagged.
• By analyzing the retweet patterns of experts, you can understand their areas of interest and spheres of influence. You can do it with a few  open source tools.
• You can understand how information propagates – what, why, how, when by analyzing tweets. Organizations like InfoChimps, Datasift can provide you with a large body of tweets you can use for research. You can make intelligent guesses based on the velocity of propagation (speed at which topics trend)
• A lot of papers have been written about Modeling Topics from Tweets. You can use several techniques to create your own watch signals in a specific space (market, industry segment, geographic region etc.)

If you are interested in this area contact me at dorait (at) imorph.com. Will be happy to answer any questions, elaborate some of these ideas and have a chat.