Gregory Maciag

Big Data v Lot's of Data

It's a small word, but it's being misused. What does the “big” in big data mean?

Figures similar to the following are coming to a slide show near you: “It has been estimated that Wal-Mart records more than 1 million customer transactions each hour, resulting in more than 2.5 petabytes of data being stored, the equivalent of 167 times the data stored in the Library of Congress. Facebook is reported to store 40 billion photographs. And an estimated 35 hours of video content is uploaded to YouTube every minute.”

This is interesting stuff, but it mixes apples and oranges. WalMart cranks out a lot of data – but it's structured data. We're talking about a lot of small things. But pictures and video are another matter. These are large, unstructured data objects – what I'd call genuine big data.

Genuine big data poses novel problems of interpretation and combination. Pseudo big data – that is, large transaction volumes – might tax the processing capabilities of traditional databases, but it doesn't challenge the analysis model in the same way as unstructured data does.

So, when someone raises the problems of big data, it's worth asking whether they're talking about high volumes of structured data, unstructured data, or both. These are different topics and they call for different strategies.  Big Data

 

Standards News from the London Market

Updates on Electronic Claims and Standardization in London.

Download ECF news Apr12PUB

Download LMG News Apr12PUB

 

Metrics for Industry Standards

Here is a copy of the most recent tech Focus column on Metrics that we published.

Download ID 2012-04 Tech Focus

Data Standards Competition

Michael Shashoua of Inside Reference Data asks a good question about competition between data standards. His focus is the financial services domain, but the question can be adapted to any other sector: “[W]hich approach is the right way to go for data standards modeling? Is it emphasizing organization or classification by asset class? Or is it linking the proverbial alphabet soup of acronym-named formats together?”

Shashoua is here contrasting the Open Data Model (ODM) with the Financial Industry Business Ontology (Fibo). He invites readers' comments, and I guess many will respond along the lines of “it depends...”

If you're modeling a piece of business as a way of understanding it, scoping a project or building a multi-disciplinary team, then the ODM's asset class view will be helpful. If you're looking to bring interoperability to processes and systems, the Fibo view will be helpful.

At ACORD we're convinced that people will always need multiple, related views. You can't force a single perspective on to all users and audiences. The ACORD Framework therefore contains different, linked models that we call facets. These enable access to the insurance domain from different viewpoints.

There's a meta-question underlying Shashoua's question, which is: Should standards compete? My answer would be that different approaches to a single standard should indeed compete, but that single standards should not themselves compete with each other. In other words, encourage debate and diversity within the standard setting process.

Data Standards Competition

Electronic Medical records

An artcile written by Chris Taylor talks about electronic medical records.  In it, he says: "There are few more personal, passionate, and political topics than health care. The reasons for this are clear: Health care spending has reached 17% of the U.S. GDP, outcomes are worse (partly due to  American lifestyle issues in my opinion) than in other developed countries, and an attempt to fix the system through the Affordable Care Act (ACA) now sits in the hands of the U.S. Supreme Court. But regardless of ACA's legal prognosis, the Pandora's Box of true health care reform has already been opened — and it happened before most of us realized."

Download The Health Care Reform That Can't Be Stopped 

Data Visualization

Do people visualize data? This is a question that's never asked because the answer is supposedly self-evident. Of course people visualize data! That's why they draw pie charts and bar graphs. Duh.

However, I think there's a limit to how far people can visualize data – and this is a human limit, not a technological one. At a certain point, visualization cannot help us any more. It concerns me that efforts designed to bring greater visualization to data might be better devoted to other activities.

What's the largest number you can comfortably visualize? I'd guess it's twelve, which you can see as the spots on a pair of dice. Anything more than a dozen or so and your brain gives you a fuzzy picture labeled “lots”.

Then there's relationships between things. Two things joined by a single relationship is pretty easy to visualize. So is a triangular relationship. After that, it gets tricky. Sure, you can visualize a polygon with any number of sides or cross-wiring between the vertices. But can you remember what each relationship means? That's not so easy.

William Gibson's cyberpunk vision of an immersive world of data was seductive, but it's not reality. You remember, in Gibson's future people jacked into the net and could then experience data as a kind of tangible environment. Today we have tools that allow us to plot all kinds of structures and graphics from data. We can fly over terrains built entirely from abstract data. The problem is, we learn nothing by it.

In fact, we learn most precisely where data visualization fails us. We realize the enormity of an effect when it's off the scale. We innovate when we think outside of the box. We have insights when our data delivers patterns we've never seen before, and which force us to ask new questions about the business.  Greater data visualization isn't going to bring significant breakthroughs. Human vision will.  Visualization

ACORD Board Aligns Plan with Vision

The ACORD Board of Directors met this month to review 2012 Plans and the 2020 Vision. The journey almost always begins with current plans and budgets.  We've been working on our agile standards development project to streamline the process.  We need to speed standards to market with a long range goal of distributing plug and play deliverables that require little if any tweaking for implementation.  Software components are in the planning stage.  We are also exploring web services as an emerging channel for the distribution and use of industry standards as well.  

We are creating a new standards development platform to replace the various databases and processes accumulated over the years from different sources. The platform will use the ACORD Framework (i.e. Models) in order to harmonize all aspects of the process.  We will be creating a client portal for use by developers for consuming standards into the enterprise. Our models and a streamlined platform will make it easier for us to focus on the distribution of implementation tools. We are planning for a tool to translate ACORD standards for example. 

An ACORD Forms portal is also in the plan to expedite updates and better inform users of agency systems.  Testing and certification will get the additional rigor required to support trading partners.  It works well in London and v2 will be expanded to include other lines of business.  We're also looking at how we can support mobile applications, some of which are already in use today.

Finally, we will continue our goal of seeking partnerships with industry organizations and continue advocacy for industry standards. Such collaboration will get us involved in eSignatures, Single Sign-on Passwords, ePolicy delivery and more.  And we have engaged a firm to work with us on both industry and organizational Metrics and plan to deliver new insights and tools for members to measure the seemingly illusive value of industry standards in large scale development projects. 

Thank you for your continued support of this industry effort.  Let me know your thoughts as your feedback will help us shape this vision to better meet your needs.

Open Data Standards???

I'm all for open standards, I'm all for data standards. I'm not so sure about the Open Data Standards initiative.

I see what they're trying to do. Set up in Finland in 2011, the Open Data Standards Association is a not-for-profit foundation whose “goal is to ease the practical exchange of information assets on the internet for any commercial or non-commercial use and independent of a target audience”.

The Association is publishing a hierarchy of definitions. I stumbled into the collection at “bungee jumping”, which I was surprised to see defined as a type of place rather than an activity. This is to do with the role it plays in a hierarchy of places differentiated by usage.

My concern is that by being “independent of a target audience”, the Open Data Standards will not serve any particular community. There is undoubtedly an audience for data standards around, say, types of publicly usable building structures, but I would say that this audience should be driving such standards.  Open Standards

Social Media: Cover the Risks

Social media seems to have taken over the world.  This statement is not hyperbole. Think Facebook. Read our OpEd from Insurance Day.  Download ID2012-03SocialMedia

 

Millennial's Take on Social Media

Jeff Yates (ACT) said that he was always looking to get the upcoming generations’ take on the technology trends.  I thought you would particularly enjoy this month’s ACT article written by Lauren Foy who is a sophomore at the University of Rhode Island.  Her article is an expanded version of an email she wrote to her dad, Mike Foy, an independent agent and active industry participant.

Download ACT Article Millennial'sTake Social Media

 

Standards Must Be Inclusive

"External operational risk data consortiums are evolving, but not all in the same direction – and significant differences of opinion remain about what consortiums should be doing. Consortiums disagree over what sort of loss events should be recorded, what information should be collected on each one, and whether data needs to be manipulated to keep it relevant to different-sized companies. And they are questioning what the fundamental purpose of external data should be – for operational risk management primarily, or for measurement and modelling as well."

Loss consortiums deal with big numbers, but that doesn't mean the concepts they deal in are necessarily complex. There are two issues in the controversy outlined in this quote. The first is that different players in an industry disagree about what needs to be measured, and how it should be measured. The second is that people are unsure of “the fundamental purpose” of data.

 Let me deal with the latter issue first. There is no “fundamental purpose” to any data. Data are designed entities that provide answers to questions. Every piece of data owes its existence to a prior enquiry. The idea that data has a life independent of such enquiries is wrong. Disputes about “fundamental purpose” must, then, really be about whose form of enquiry takes precedence in the design of a data scheme.

This takes us neatly back to the first issue – the what and how of data. Now, if consortiums have differences of opinion over what information needs to be recorded, this suggests first that their business models consume and/or produce information in different ways, and second that they need to share data. (Otherwise, why would they be arguing? You hold your data, I hold mine – who would care about whether or not we were compatible?)

So, in the interests of trade, any standard needs to embrace every possible usage of a data item. Standards have to be inclusive. It makes no sense to trim standards, even though the further you get to the periphery of the standards set, the fewer users you will find. If you fail to serve the periphery, you lose trade at the core.

Think of it this way. Say we're building a shopping mall. We agree quickly that we need a parking lot and elevators. Some people in the management committee want there to be special parent-and-child parking slots. This is going to make the parking lot a little more expensive. They'd also like wider elevators, so that moms can get their double strollers in – that's going to cost a lot more.

Do you abandon this section of the customer community just because their needs are different from those of the majority? They're going to spend money in your mall. They might be the best customers your mall has.

To summarize, I say that standards bodies are well advised to take the time to be inclusive, rather than waste energy arguing about what should be excluded. I also say that data has no “fundamental purpose”. People have purposes – data has destinations.

Loss Data Consortiums