One of the benefits of the disruptive nature of emerging big data platform technologies is that the combination of scalable performance and lowered costs for high-speed memory opens the door for addressing business problems in ways that used to be too computationally-intensive to roll out on a broad scale. One good example is graph analytics, which is an area of analysis that focuses on identifying actionable knowledge inherent in the relationships between and among entities, as opposed to typical analyses that focus on characterizing individual entities.

It is simplistic to say that we live in a connected world, especially as our environment is increasingly engineered to connect people and things together. Interstate highways, the national rail networks, telecommunications networks, public utilities and power grids are all examples of infrastructure designed to link people, places and things. And although infrastructure is typically viewed as an “under-the-hood” necessity, from an intelligence perspective, the relevance of the interconnectivity lies in analyzing the way that infrastructure is used, such as which individuals connect to web sites using their smart phones, where delivery trucks make stops along a particular route on a daily basis, or how frequently individuals post to each other’s blog entries.

In the real world, entities might be connected in a number of different ways. These links are manifested in different ways, with different levels of significance, and they may extend beyond connecting entities of the same type. Interestingly, the link itself embodies characteristics and features of the connections. Here are some examples:

• One person is married to a different person.
• A delivery truck picks up packages from a residential location.
• A residential location uses power provided by an energy utility.
• One telephone makes calls to another telephone number.
• A physician prescribes a particular medication.

There are two obvious facets of each of these examples: first, each establishes a relationship between two entities, and second, the relationship itself can be attributed with magnitudes and other characteristics, such as the number of years that two people have been married, the number of packages delivered, how much power is consumed, whether a call connected residential or business accounts, or the number of times that physician has prescribed the medication.

And there is at least one inherent feature as well, which is that connections of different types can be overlaid on top of the same sets of entities. For example, when perusing the connectivity model for residential locations, one might see patterns in which some locations are frequently visited by delivery trucks picking up large numbers of packages and have a relatively high power usage and are associated with a telephone number that makes many calls to telephone numbers associated with business locations and draw the conclusion that these may be locations in which a person might be operating a business out of a residential home.

Over the next few postings we will drill deeper into the concept graph analytics and look to understand what graph models are and what kinds of information we can derive from graphs and connections. We'll also look at some examples showing the value of graph analytics.