See also: Keywords are going going, as graphing Entities comes in
[webmasterworld.com] for another comment on the use of graphing analytics/DB.
A couple of astounding numbers for those already feeling overwhelmed:
* in the western developed world there are currently, depending on country, an average of 8-12 connected devices, i.e. smart TVs (fastest growing segment), smart phones (plateauing segment), tablets, laptops, per household (IAB).
* by 2020:
---there are expected to be between 20 and 30 billion with a B devices connected to the internet (McKinsey).
---IoT annual data will exceed 600 ZettaBytes aka 600 sextillion bytes aka 600 * 2^70 bytes (Cisco).
Note: can you say GIGO?
So there will be a great great great many interconnecting relationships to be (or perhaps better put as attempted to be) understood. If you thought your personal relationships were difficult... your data relationships are a whole other universe of hurt.
Graph databases/analytics are already being used in an increasing number of verticals for an increasing number of reasons. Several uses cross many/most lines, i.e. fraud detection, regulatory compliance, supply chain transparency... and the one I'm speaking to today: connecting the user dots aka xdevice (cross device) aka multiple screen (marketing double speak) usage.
So I visit your site from a smartphone while on transit in the morning, from a laptop at work during lunch, from a tablet at home that evening... if there is no registration/login as with WebmasterWorld how do you know that it's me? Three different devices, three different locations, three different time periods. If you use session cookies there are three different of those as well.
It won't be too big a surprise to learn that much of the effort in this dot connecting is, and has been, driven by marketing and advertising requirements. FB's 'do you know' is graph driven as is Amazon's 'you may also like'. But the current race is on reliably connecting (1) each of us with each of our connected devices and (2) our relationships with others at same locations or using same devices.
A common misconception is that the EU's GDPR and upcoming ePrivacy regulations will drain the data ocean that is prevalent and so abused. It will curtail some behaviours in some areas. However, let's be honest (beware, watch for con ahead!): most sites don't actually need to know all your personal details, an anonymised ID in place of name, address, and phone number is often just as valuable.
Why? Whether a site relationship is with John Doe of 123 Main Street, Someplace, Some Postal Code, 999-555-1000 or with E02799F94915229B13B5A73FC280CDF6 really doesn't matter except when it comes to payment and shipping. A rose by any other name would smell as sweet. The one big, and it truly is big, difference is that hashed ID data becomes pretty much useless for sale to third parties (until...). However, within a site (or pooled sites) identifying 'you' and contextually delivering content, for all practicality, nothing/little changes. The site can, in essence, just as easily pick up the 'conversation' where it left off.
Yes, I know that many/most of you (although perhaps not those few who have read this far) are strictly SE aka Google one time and gone visitors; for you the relationship (SEO or S&M?) is with the referrer not the visitor, you manically speed date each visitor hoping for instant gratification/conversion as there is little/no second chance.
However, for those who work to optimise traffic referrers other than search and build for return visitors, actually knowing which are first timers and which are returnees is not an easy challenge that is made increasingly difficult by multiple devices by single users. How best to run a coherent marketing campaign across such divides? Now throw in simultaneous multiple users at same location on various devices, i.e. all checking up on something in a TV program or Netflix movie.
While marketing is a/the main impetus another piece of the analytics pie where understanding relations can be valuable is in bot detection. Actually, the deeper I go down the graph analytics/database hole, especially tied in with ML (both as input and output), the more use cases I find.
In a similar, OMG! what have I got myself into, scenario: after 41 years Voyager 2 has exited the heliosphere and entered interstellar space following Voyager 1, which did so in 2012.
To boldly go...
For fools rush in...