Complex Insight - Understanding our world

Platfora has gained a lot of buzz in the Big Data analytics market primarily through word of mouth. Late last year the company...

Results from Kalido survey  in London – Big Data Analytics 2012. Key take aways: Only 15% had a Big Data Analytics project underway and another 37% planned to do something within 1 year. However, 26% have no timeline in place.  This maps with our internal surveys where we are seeing companies with a lot of machine based data (log files, call data records and the like) looking at the technologies while general business is adopting a wait and see how this thing matures, pans out and is relevant.  Click on the image or the title to learn more.

Vincent Huang has a great set of posts at Ericsson labs on preserving privacy in big data analytics. Over the last few years there has been a growing trend in the international development world to highlight the use of big data for development. Using mobile telephony call records, twitter postings etc as raw data sets that can be mined for indicators of social circumstances and systemic change. A key first step in this process is anonymizing the data so individuals cannot be easily identified from distilled results. Vincent's work discusses the requirements and algorithms used and is both a fascinating and essential read for anyone working in this area. Must read posts if you are involved in big data for development or working with private individual identifiable data sets. Click on the image or title to learn more.

Titan is the latest supercomputer to be deployed at Oak Ridge National Laboratory (ORNL), although it's technically a significant upgrade rather than a brand new installation. Jaguar, the supercomputer being upgraded, featured 18,688 compute nodes - each with a 12-core AMD Opteron CPU. Titan takes the Jaguar base, maintaining the same number of compute nodes, but moves to 16-core Opteron CPUs paired with an NVIDIA Kepler K20 GPU per node. The result is 18,688 CPUs and 18,688 GPUs, all networked together to make a supercomputer that should be capable of landing at or near the top of the TOP500 list. Click on the image or the title to learn more

Interesting article on CMS wire regarding insight and big data. What is the Big Data Fallacy? Data does provide information, and more data generally gives more information. However, the fallacy of big data is that more data doesn’t lead to proportionately more information. In fact, the more data you have, the less information you gain as a proportion of the data. The return on extractable information from any amount of big data asymptotically diminishes as your data volume increases. In nearly all realistic data sets (especially big data), the amount of information one can extract will be a very tiny fraction of the data volume: information << data. What about insights? How does that relate to information and data? All insights are information, but not all information provides insight. In order to provide insight, information must be: Interpretable, Relevant, Novel. Click on the image or the title to learn more.

Health care in the United States certainly needs an overhaul. The question is whether that overhaul will come from artificially intelligent doctors. IBM  has launched partnerships with insurance giant WellPoint and the Sloan-Kettering Cancer Center in New York and is expected offer Watson commercially to hospitals within the next few years. Wired article discusses the approach and pros and cons as seen by some doctors. Worth a read for a quick intro to modern AI comprising data mining and big data topics as an approach to treatment diagnostics in healthcare. Click on the image or the title to learn more.

Get into a cab and it's safe to assume the driver knows the ins, outs, shortcuts and potential traffic tie-ups between you and your destination. That kind of knowledge comes from years of experience, and IBM is taking a similar tact that blends real-time data and historical information into a new breed of traffic prediction.

IBM is testing the new traffic-management technology in a pilot program in Lyon, France, that’s designed to provide the city’s transportation engineers with “real-time decision support” so they can proactively reduce congestion. Called Decision Support System Optimizer (DSSO), the technology uses IBM’s Data Expansion Algorithm to combine old and new data to predict future traffic flow. Over time the system “learns” from successful outcomes to fine-tune future recommendations. Click on the title or image to learn more.

Thanks to Jed for spotting this. Barry Devlin's blog offers a good insight into the emerging discusions around unified data management platforms that can scale from small to big data architectures. Barry's recent white paper on Data Zoo's (linked from article) expands on the points made. Both are worth a read. Click on the image or the title for more info. 

Over the last year we have been experimenting with different data processing platforms including Hadoop, HPCC, several graph databases and NOSQL databases. WE selected a hybrid of STORM and HPCC since combined they are a  great match to our processing needs, both have permissive open source licenses and good resources.

Many Big data articles fail to mention that Big Data applications have varying requirements and the different available toolsets suit different applications and circumstances. One of the reasons we went with HPCC is we have good C++ skills on our team, we liked the ECL language and systems architecture. We could see how pur target workflows would map (it helps that the Lexis Nexis team have addressed similar areas int he past and it shows in the product) and we could easily see how to integrate the HPCC functionality with our C++, Java and Python tools.

If we were a pure java only house then HADOOP,CASCADE and HIVE would be a much better match for non real time procesing. Tools like Storm are better architected for real time event processing (hence our hybrid choice).  One of the essential lessons and take aways is that each domain and circumstance has different requirements. You have to play with the technologies, test them against workflows and see what works for your problem domain, performance requirements and team skills.  For evaluating HPCC a great resource is Arjuna's blog. You will find worked examples discussion of trade offs and approaches and a lot of good information. If you are interested in learning about a great Big Data platform - its a good place to start. Click on the image or the title to learn more.

Wired has an interesting article on mobile computing. As ARM seeks to put cell phone chips into our supercomputers, Intel is doing the reverse. The lines between the mobile hardware and data center hardware are blurring. Intel’s experimental Single-chip Cloud Computer project, or SCC is a 48 core chip acts as a “network” of processors on a single chip, with two cores per node. The nodes actually communicate to each other much the same way nodes in a cluster in a data center would.  While these types of architectures are today used in big data applications that run on a server cloud environment there are actually many cases where a hybrid model would make the most sense including artificial intelligence, machine vision and augmented reality applications.  Click on the image or the title to learn more.