Wednesday, April 24, 2013

Unwind the Mind : Data Dynamics of Natural Computers

Data Diary (5): Natural Computers

Information space continues to enrich my imagination to new manifolds. The recent news that a new start up named Aysadi has come up with a Topological approach to machine learning and Big Data analytics, is really really an interesting conjecture. It is said that DARPA, NSF and Stanford were involved in this research for a long time. I hope this will be a good trend where we will have an open approach to data science. This should be beyond the dependency on specific tools. 

This inspired me to go through some nuances of topological learning and created lots and lots of questions in my 'unstructured' mathematical understanding. The stress on various 'in variance' conditions in topological analysis makes me believe that we are far from the best approach. A comprehensive approach to a data problem should not be defining a boundary to its explorations and insights. Yet my comment remains largely naive as I am not an authority or trained in topology. 

Continuing from our previous post on information - cognition   conjecture, I have landed on a cyclical condition. With the advent and advance of cognitive computing and neuroscience, we are creating anew computing machines driven by human cognition. So we can state that cognition can control computation and therefore information too. On the other side of the coin, can information control cognition. In simple terms the answer is yes, a plain yes. If so, can we create a cyclical information - cognition cyclical machine ? This should be a machine where cognition initiates information processing and then information processing generates new re-cognition. 

When I try to rationalize this order, I believe this is happening in all our day to day lively transactions. Going on the same lines, how many machines can claim to do this natural computing cycle to maximum approximation to the real world. And what is the most effective model to observe the data flow in this cognition - information - re-cognition cycle. Knowledge ( Neural Signals, Thought Processes ) in ( Cognition ) - (Language, Semantics, Syntax) in ( Information ) - ( Semiotics, Visuals, Shapes, Numbers, Senses, Emotions ) in Re-cognition seems to be data dynamics. Natural computing demands more rigorous modelling for data dynamics. In pursuit of more natural thoughts ... 

PS: Content is Social. Social is Me

Monday, April 22, 2013

Data Diary #5: Hyper Cubes of Information spaces

When, What and Where is Content ! 
3Ws of Data Science ...

When I was working on a meta data strategy for a data governance initiative, I came across the below interesting point :  
When to locate content , 
What is content and where to locate the content.

I believe these three thoughts works behind many of the search engine driven meta data strategies. They I came across a data dilemma  why do we see metadata becoming stale and stealth? And it made me to think about information spaces, their temporal properties and how to visualize them. Are they like the conventional space time conjectures and curvatures ? Then I realized that space-time is never an absolute metric of anything. 

Let's take a few information spaces. One imminent example that comes up in our mind is a library of wealth of information. Another one can be a stock market where numbers and stocks flock with finance capital. Yet another one can be a group of people assembled in a parliament or a conference.  And a very familiar example of a convenient information space is a data warehouse or a relational database. This is largely a information space sans soul of information. 

Connected Spaces
All these are information spaces and they define their metrics of content and metadata. Often information space is just associated with the needs of data visualization. This approach will provide only limited perspective of information spaces. Each information space is having a temporal or contextual aspect embedded or evolving around it. And it cannot be simplified in some relations of data structures. If data structures should meet this criteria, they should have a time variant structure associated with them.

Am I again going back to the traditional space-time? No, rather, just highlighting the necessity to accommodate  time in this situation. Every measurement system should know what it is going to measure. If this factor is not understood well, we will always witness an uncertainty or probability or chaos in measurement and results.

So if we design an information space to visualize content and metadata that locate them (When, What, Where), we need to know that all these co-ordinates themselves are manifestations of some other information spaces. Hence Information space cannot exist in silos. It exist through #Macro Connectors. The concept of macro connectors is not mine. This is proposed by MIT Media Lab. And it looks interesting. Information space thus becomes a connected space. 

So far so good. What do we achieve by extending these connections? How will information spaces work in cognitive computing / social computing environment.  Like light getting bend by gravity, I would love to say that information spaces get truncated and twisted, curled, diverged, converged by the real-time decisions of cognitive data nodes. Information overloading is just a behavior out of million possibilities in this information - cognition conjecture. 

Google: +Gokul Alex 
Twitter: @gokulgaze

PS: Content is social, Social is Me

Sunday, April 21, 2013

Data Diary #4

Information Machines and Metadata Strategy: Some early thoughts

How much of engineering is required for designing / evolving +information machines so that they can always differentiate between data and +metadata and further go beyond to create knowledge and insights out of it. While thinking of this question. I cam across the role played by Search Engines in this. No doubt, search engines are information machines. They learn the #semantic graphs of information relationships in a sea of indexes. Let me bring some metrics here. How will we measure the effectiveness of search engines as information machines. Do we need some axis to plot / visualize their effectiveness.

One of my favorite questions will be how much of data / information / relationships / indexes / semantic webs can be converted to working knowledge by Search engines? Rather can they do this task at all without the intervention of human interactions, at all ! 

Going by the same lines, my categorization / differential positioning of +metadata with respect to data will not be merely based on the relationship between the meaning and associative positioning of data nodes. It will be rather based on how one node of data connects the other node of data to a third node of data. So data and +metadata should always have more than one connecting dot between them. Thus when we use a search engine to find all the +metadata about a particular data node, it should find the meta data based on this semantic graph. 

@gokulgaze, PS: Views are my own and do not subscribe to any organization or institution

Monday, April 15, 2013

Data Diary #3:

Network Equipment Providers's (NEP) and Next Generation Content

We have already witnessed the wonders of streaming media and video streams. Now comes the real time video streams. At the recently concluded Mobile World Congress, Peter Linder,VP, Fixed Broadband and Convergence, Ericsson speaks about how much Ericsson is enthusiastic about the possibility of live and multimedia content in mega sporting events such as FIFA World Cup, Olympics and so on.

Ericsson Speaks on Next Generation Content

This speech is a pointer to the direction in which Network Equipment Providers must be moving. This is nothing but out of the realization that Voice communication networks can no longer be the only revenue generator or business value accelerator. Voice -Data convergence, Content explosion and the entrance of traditional NEP's in this ecosystem opens up interesting possibilities of competition in the content aggregation and content analytics space.

Already we have a lot of players in the content aggregation and content analytics space. How can NEP's leverage their experience and infrastructure to create new opportunities and user experience ? And that is something worth study.

+Gokul Alex