Thanks to #SITNL and Jan Penninkhof I learned about a solution by SynerScope. These guys have created a great data visualization tool using saphana for their data. The tool itself is intended to visualize Network diagrams. the type of diagrams where you have different nodes, with different relations and the more nodes you add, the bigger your network hairball gets cluttered.
So these guys created a fancy tool to cluster relationships, and visualize them on a timescale. As you select a timeslice, the network diagram updates along with other possible visualizations (such as scatter diagrams, geomaps, trend graphs etc..)
If someone would explain me their solution like that, I would still say: Yeah, so? Until they give me an example.
The example used by the Synerscope guys, was an analysis of a Botnet. WHoaaa, say what?
Yeah, I never imagined using SAP HANA, or a BI tool to analyse Network security threats. But if you think about it, it makes sense. The relationships between servers and nodes in a botnet are massive. The only way to analyse and find the roots is by going through all the traffic data, mapping it out on a geomap and in a network graph, and going back through time to find the root cause.
That’s exactly what they did, and as they reduced their timeslice you could really see the network folding in on itself to a single starting point. Fast forward and you see how the network unfolds and new servers get active and infect nodes. Pretty darn cool.
Take a look at the case study on their website to get an idea.
Other use case
So I started thinking of the possibilities with this tool. There are so many real-life examples of networks to analyse, and there is a lot of information to be mined doing so.
Public transport obviously has a vast network of railways (relations) and stations (nodes). And that is only for trains. Take the buses as an example, and you have an even bigger network of lines and stops. Or take air traffic with all the connections and airports. Yeah, you get the idea, public transport is one giant network. Now if you can analyse this network to find the most effective route between two points (least nr of hops and least time wasted) you can start optimizing your schedule.
I know that in Belgium, the railways get a lot of comment each year, when they present their new roster, and it’s perceived as less effective each time. Publish the roster with such a visualization (interactive tool/online maybe) and people can see for themselves whether it’s better or not.
The social universe is one big cluster of relations and nodes. Every person is connected to many other people. Take LinkedIn, Twitter, Facebook, Jam,… If you can represent these relations and actively mine for hubs, people with a vast reach, you can actively market influencers who will forward the message. Nowadays, marketeers typically look at the number of 1st line connections, but there might be people out there with only 200 direct connection, but with a reach that expands much further, because their connections have many more connections in turn.
By doing hop degree analysis, you can find the hidden gems in the Socialosphere.
In the movies, when a hacker steals a lot of money, they run a program to distribute all the money over many banks and make it move all over the world until no-one knows anymore where it is, and then the hacker makes it converge on the Cayman islands and lives in a lodge on the beach (
remember Swordfish?). That’s probably not reality, but even so, by doing such timesliced analysis on financial transactions, banks could actually track that money anywhere.
This is a no-brainer and very similar to the public transport case. Identify all your warehouses, customers and delivery routes, and find out if you can optimize any by doing a time analysis. Combine it with navigation map data, and you can even come up with alternative routes to your customers.
My first thought when seeing this was: Likes and Ratings on SCN! Load the people and the ratings into such a tool, and the network graph will tell you where there are unusually high activities. On the other hand, it will also show you who is actively promoting other people’s content, and who is receiving a lot of promotion. this can be used in a positive way to identify quality, or in a reactive way to identify spoofing.
I worked in the Utility sector for more than 5 years, and still, this was the last thing that crossed my mind when seeing the demo. But actually, it makes a lot of sense. Utility networks are huge, and must be monitored constantly. You need to know where the water or electricity is flowing, which alternative routes exist in the least number of hops possible if a node goes down and where there is an unusually high load.
I’m pretty sure there are lot’s more use-cases, and that SynerScope has thought already of many of them.
Still, it doesn’t hurt to suggest some ideas and give the guys a bit of publicity. I think they have an interesting product and that there are many customers that can benefit from it.
Can you think of any other use cases to analyse network Graphs?
(and I’m looking at some of the sitnl participants who already raised there voice during the demo.)
Grid image from Flickr: http://www.flickr.com/photos/jprovost/5532276866/sizes/m/in/photostream/
Money image from Flickr: http://www.flickr.com/photos/amagill/3367543296/
Synerscope image from: http://www.synerscope.com