Here is a great call for Open Linked Data by Ian Davis from Talis:
Conjecture 1: Data outlasts code , therefore open data is more important than open source.
Conjecture 2: There is more structured data in the world than unstructured, therefore people who understand structure matter.
Conjecture 3: Most of the value in our data is unexpected and unintended, therefore we should engineer for serendipity
I am not sure about Conjecture 2. I thought there was more semi-structured and unstrctured data than structured data, but I could be wrong about that.
So what does it take to discover the unexpected and uninteded value of data? Data Mining with machine learning algorithms?
@Dorai, interesting points. I agree with your comment on Conjecture 2. There is way more semi or loosely structured data in the world.
As for your question on discovering the unexpected value of data – I find this approach helpful. Imagine a graph with your current standpoint situated in a specific cluster of nodes. Now any nodes that are connected to you by a single edge are ‘known’ and ‘obvious’. Do 2 hops or more and you have reached areas that were opaque to your perspective earlier. I use this to imagine scenarios that lead to unexpected insights.
Whether this is done with DM and ML is an implementation details I would think.
Thanks for the comments. I think DM (which may include) may even discover those edges which may not be explicitly described are specified (from semi/unstructured text, I guess that is what the semantic miners have to do).
It is nice to chat with some one with similar interests. I follow your tweets and the FF entries and find them very useful. Thanks for sharing all that info. Will you be at BCB8 (barcamp at Bangalore) on 7th/8th of March? I would love to have the chance to meet you and swap stories.