This post is about Big Data. We will talk about the value and economical benefits of Big Data, not the atoms that constitute it [Big Data]. For the atoms you can refer to Wearable Technology or Getting Ready for the Internet of Things by Alex Sukholeyster, or just logging of the clickstream… and you will get plenty of data, but it will be low-level, atom level, not much useful.
The value starts at the higher levels, when we use social connections of the people, understand their interests and consumptions, know their movement, predict their intentions, and link it all together semantically. In other words, we are talking about six graphs: Social, Interest, Consumption, Intention, Mobile and Knowledge. Forbes mentions five of them in Strategic Big Data insight. Gartner provided the report “The Competitive Dynamics of the Consumer Web: Five Graphs Deliver a Sustainable Advantage”, it is a paid resource, unfortunately. It would be fine to look inside, but we can move forward with our vision, then compare to Gartner’s and analyze the commonality and variability. I foresee that our vision is wider and more consistent!
This is mostly analyzed and discussed graph. It is about connections between people. There are fundamental researches about it, like Six degrees of separation. Since LiveJournal times (since 1999), the Social Graph concept has been widely adopted and implemented. Facebook and its predecessors for non-professionals, LinkedIn mainly for professionals, and then others such as Twitter, Pinterest. There is a good overview of Social Graph Concepts and Issues on ReadWrite. There is good practical review of the social graph by one of its pioneers, Brad Fitzpatrick, called Thoughts on the Social Graph. Mainly he reports a problem of the absence of a single graph that is comprehensive and decentralized. It is a pain for integrations because of all those heterogeneous authentications and “walled garden” related issues.
Regarding the implementation of the Social Graph, there are advises from the successful implementers, such as Pinterest. The official Pinterest engineering blog revealed how to Build a Follower Model from scratch. We can look at the same thing [Social Graph] from a totally different perspective – technology. The modern technology provider Redis features a tutorial on how to Build a Twitter clone in PHP and (of course) Redis. So the situation with Social Graph is less or more established. Many build it, but nobody solved the problem of having a single consistent independent graph (probably built from other graphs).
It is a representation of the specific things in which an individual is interested. Read more about Interest Graph on Wikipedia. This is the next hot graph after the social. Indeed, the Interest Graph complements the Social one. Social Commerce sees the Interest + Social Graphs together. People provide the raw data on their public and private profiles. Crawling and parsing of that data, plus special analysis is capable of building the Interest Graph for each of you. Gravity Labs created a special technology for building the Interest Graph. They call it Interest Graph Builder. There is an overview (follow the previous link) and a demo. There are ontologies, entities, entity matching, etc. Interesting insight about the Future of Interest Graph is authored by Pinterest’s head of engineering. The idea is to improve Amazon’s recommendation engine, based on the classifiers (via pins). Pinterest knows the reasoning, “why” users pinned something, while Amazon doesn’t know. We are approaching the Intention Graph.
Not much could be said about intentions. It is about what we do and why we do. Social and Interests are static in comparison to Intentions. This is related to prescriptive analytics, because it deals with the reasoning and motivation, “why” it happens or will happen. It seems that other graphs together could reveal much more about intentions, than trying to figure them [Intentions] out separately.
Intention Graph is tightly bound to personal experience or personal UX. It was foreseen in far 1999, by Harvard Business Review, as Experience Economy. Many years were spent, but not much implemented towards personal UX. We still don’t stage a personal ad hoc experience from goods and services exclusively for each user. I predict that Social + Interest + Consumption + Mobile graphs will allow us to build useful Intention Graph and achieve capabilities to build/deliver individual experiences. When the individual is within the service, then we are ready to predict some intentions, but it is true when Service Design was done properly.
One of the most important graphs of Big Data. Some call it Payment Graph. But Consumption is a better name because we can consume without payment, Consumption Graph is relatively easy for e-commerce giants, like Amazon and eBay, but tricky for 3rd parties, like you. What if you want to know what the user consumes? There are no sources of such information. Both Amazon and eBay are “walled gardens”. Each tracks what you do (browse, buy, put into wish list, etc.), how you do it (when logging in, how long staying within, sequence of your activities, etc.), they send you some notifications/suggestions and measure how do you react, and many other tricks how to handle descriptive, predictive and prescriptive analytics. But what if a user buys from other e-stores? There is the same problem as with Social Graph. IMHO there should be a mechanism to grab the user’s Consumption Graph from sub-graphs (if the user identifies herself).
Well, but there is still big portion of retail consumption. How to they build your Consumption Graph? Very easy, via loyalty cards. You think about discounts by using those cards, while retailers think about your Consumption Graph and predict what to do with all of the users/clients together and even individually. There is the same problem of disconnected Consumption Graphs as in e-commerce because each store has its own card. There are aggregators like Key Ring. Theoretically, they simplify the life of the consumer by shielding her from all those cards. But in reality, the back-end logic is able to build a bigger Consumption Graph for retail consumption! Another aspect: consumption of goods vs. consumption of services and experiences, is there a difference? What is the difference between hard goods and digital goods? There are other cool things about retail, like tracking clients and detecting their sex and age. It is all becoming the Consumption Graph. Think about that yourself:)
Anyway, Consumption Graph is very interesting, because we are digitizing this World. We are printing digital goods on 3D printers. So far the shape and look & feel are identical to the cloned product (e.g. cup), but the internals are different. As soon as 3D printer will be able to reconstruct the crystal structure, it will be a brand new way of consumption. It is thrilling and wide topic, hence I am going to discuss it separately. Keep in touch to not miss it.
This graph is built from mobile data. It does not mean the data comes from mobile phones. Today may be the majority of data is still generated by smartphones, but tomorrow it will not be the truth. Check out Wearable Technology to figure out why. A second important notion is about the views of the understanding of the Mobile Graph. Marketing based view described on Floatpoint is indeed about smartphone usage. It is considered that Mobile Graph is a map of interactions (with contexts how people interact) such as Web, social apps/bookmarks/sharing, native apps, GPS and location/checkins, NFC, digital wallets, media authoring, pull/push notifications. I would view the Mobile Graph as a user-in-motion. Where user resides at each moment (home, office, on the way, school, hospital, store, etc.), how user relocates (fast by car, slow by bike, very slow by feet; or uniformly or not, e.g. via public transport), how user behaves on each location (static, dynamic, mixed), what other users’ motions take place around (who else traveled the same route, or who also reside in the same location for that time slot) and so on. I am looking at the Motion Graph more as to the Mesh Network.
Why dynamic networking view makes more sense? Consider users as people and machines. Recall about IoT and M2M. Recall the initiatives by Ford and Nokia for resolving the gridlock problems in real-time. Mobile Graphs is better related to motion, mobility, i.e. to the essence of the word “mobile”. If we consider it from a motion point of view and add/extend with the marketing point of view, we will get a pretty useful model for the user and society. Mobile Graph is not for oneself. At least it is more efficient for many than for one.
This is a monster one. It is about the semantics between all digital and physical things. Why Google rocks still? Because they built the Knowledge Graph. You can see it in action here. Check out interesting tips & tricks here. Google’s Knowledge Graph is a tool to find the UnGoogleable. There is a post on Blumenthals that Google’s Local Graph is much better than Knowledge, but this probably will be eliminated with time. IMHO their Knowledge Graph is being taught iteratively.
As Larry Page said many times, Google is not a search engine or ads engine, but the company that is building the Artificial Intelligence. Ray Kurzweil joined Google to simulate the human brain and recreate the kind of intelligence. Here is a nice article How Larry Page and Knowledge Graph helped to seduce Ray Kurzweil to join Google. “The Knowledge Graph knows that Santa Cruz is a place and that this list of places is related to Santa Cruz”.
We can look at those graphs together. Social will be in the middle because we (people) like to be in the center of the Universe:) The Knowledge Graph could be considered as meta-graph, penetrating all other graphs, or as super-graph, including multiple parts from other graphs. Even now, the Knowledge Graph is capable of handling dynamics (e.g. flight status).
There are other graphs in the world of Big Data. The technology ecosystems are emerging around those graphs. The boost is expected from the Biotech. There is plenty of gene data, but lack of structured information on top of it. Brand new models (graphs) to emerge, with ease of understanding those terabytes of data. Circos was invented in the field of genomic data, to simplify the understanding of data via visualization. More experiments could be found on Visual Complexity web site. We are living in a different World than a decade ago. And it is exciting. Just plan your strategies correspondingly. Consider Big Data strategically.