This page was translated from French to English with the help of AI. Read the original in French.
The rest of my talk at Que Du Web 2017 in Deauville.
It’s about graphs, history (with a small h and a capital H), algorithms and practical applications, SEO in particular.
In this second episode, I try to convince you that trees and graphs are actually part of our history, and have always been used to organize and pass on knowledge.
(See the overview of the talk with all the episodes)
The matching slides:
The transcript
In episode 1, we saw that graphs are everywhere, including in everyday life.
And I was telling you that the history of graphs, of trees, goes back even further…
To try to convince you, we’re going to take a few trips back and forth through time together.
Here’s a modern tool, “Answer the Public”, which shows the questions associated with a term, here “graph”, as a tree.
It’s a common representation when you’re trying to define personas or when you make mind maps. Did I say modern?
Actually, it’s a very old representation, and a fairly common one.
For example, I picked this tree of nature and logic, by Llull, which contains, here on the leaves on the right, 10 types of questions. You’ll recognize the definition of a persona, the questions from Answer the Public.
But also, in the center, a Porphyrian tree, which shows, in a structured way, the relationships between different concepts.
And here we’re with the Greeks, in the 3rd century. We’re talking ontology, hierarchy, tree structure, and we could go all the way back to Aristotle, in 350 BC.
Speaking of tree structures, in SEO you may know that the concepts of Trust Flow and Topical PageRank are based on a tree of categories originally defined by the DMOZ directory: categories defined and fed by humans.
Once again, nothing new.
In 1780, in Diderot and d’Alembert’s Encyclopédie, the table of contents is nothing other than a tree, which represents the different branches of science in a structured, hierarchical form.
When I say Encyclopedia, you probably think of Wikipedia before you think of the historical version. Fine.
Here’s Wikipedia.
A phenomenal set of interconnected pages, a very complex graph.
But if you look at the big picture, in terms of clusters, here’s what you get:
With a little legend to dot the i’s and cross the clusters…
we end up with a visual very close to the Encyclopédie’s table of contents.
And it holds true at different levels, here for example to show what revolves around transhumanism.
Thanks to the graph, we can see things, relationships that would be hard to take in, to synthesize, by reading the 23,000 pages.
So graphs, and trees in particular, have always been used to represent, organize and pass on knowledge.
Here’s another historical example, Llull again.
And I’m showing you this one because it’s the tree of knowledge.
And from the tree of knowledge to the knowledge graph, Google’s Knowledge Graph, it’s only a small step.
It’s a graph used to define relationships between named entities: dates, events, places, people…
Here, Ramon Llull’s Knowledge Graph.
It was a minor revolution when Google introduced it in 2012. New, really?
Yes and no. Here’s what can be considered the ancestor of the Knowledge Graph.
A diagram from the year 1000, which brings together the key concepts of the time, with various relationships between them.
In particular, you’ll find the 4 elements, which we’ll run into again later: in a moment I’ll be telling you about Euler and water (in French, “d’Euler et de l’eau” sounds like “air and water”), which gave birth to graph theory.
Now, in case you had any doubts, I’ve been talking about nothing but SEO from the start.
Whether from the point of view of the user, the search engine or the SEO, the concepts of tree and graph are at the heart of the problem.
The apparent “novelties” in SEO, the KG, intent and personas, the semantic cocoon (cocon sémantique), user paths, mind maps, content tree structure, Topical TF…
These are actually concepts with roots that go back a thousand years.
At the bottom of it, there’s often a tree. Or, when the thing you’re studying is more complex, a graph.
Here’s a more complex example, as it happens: a graph built from a dataset of 200k tweets provided by Aysun
(@AysunAkarsu, interactive link coming soon)
The ones who RT everybody.
The stars.
And the ones with a very dense community (or a bot network).
To build this second graph, I only kept the hashtags, which are therefore the nodes of the graph, and I linked them whenever they were mentioned in the same tweet.
Then, a PageRank-type algorithm to bring out the most important terms.
In “real life”, it’s an interactive graph, but there are already some interesting things to note: for example, for French tweets, “Google” is here, it’s the biggest, above everything else.
For English speakers, what carries the most weight are concepts like webmarketing and growthhacking. “Google” is here, rather off to the side and less important.
The analysis is basic, but the point is to show that even on fairly complex graphs, you can see things that wouldn’t always be obvious without them.
I hope that by now I’ve convinced you of one thing:
graphs and trees, as visualizations, are essential for structuring information, and for passing on complex information reliably.
If you dig into neuroscience, you’ll actually find reasons for this, related to the way our brain perceives and models the world.
And our brain, as it happens, is a graph of interconnected neurons.
This graph, however, is too complex for the raw visualization to be of any use.
If I showed you a graph of your brain, you’d have no idea how it works.
You could say the same about a complex website, for example.
So, in cases like that, what do we do? Are graphs useless past a certain level of complexity?
That’s what we’ll see in the next episode…