The Intent Graph
Unrivalled customer understanding
Our Intent Graph is the data enrichment engine behind Intent. By understanding the relationships between most topics in the world, we can make sense of your customers’ actions and reveal true interests.
The Intent Graph encodes an entire internet’s worth of data: subjects, ideas, people and things – more than 5 million topics. Our unique algorithms then define and link each topic on an ontological, semantic, affinity, and categorical basis, resulting in a rich web of understanding.
Ontological: the theory behind, and abstract nature of, what something is
Semantic: what something means in common understanding
Affinity: the attraction or closeness between entities or ideas
Categorical: an orderly grouping (of concepts and/or objects, topics, things, persons) compiled as a unit; including subclasses
Our ontological understanding
When we say “House” do we mean a place of residence, the verb, the TV show or the music genre? To provide the richest level of customer insight for our clients it’s important to understand the context and meaning of the data we’re dealing with.
The source of our ontological understanding is Wikidata, a collaborative knowledge network. Wikidata links the ontological DNA of concepts, objects, words and topics so we can view any natural connections and relationships at a simplified level.
Online content often covers multiple concepts and ideas under one title, so algorithms are deployed to differentiate themes and rank the importance of everything being discussed.
Our semantic understanding
For a semantic understanding, we turn to Wikipedia, quite simply because it’s human-made and all its links are made by the minds of real people.
For example, we’d find a strong semantic similarity between Usain Bolt and Asafa Powell: both are Jamaican athletes who compete at the highest level. Usain Bolt is also similar to Carl Lewis or Jackie Joyner-Kersee, as all dominated athletics in their time.
Connections are more delicate between Usain Bolt and Richard Branson but the pair’s advertising connections mean there are stronger links here compared to Usain Bolt and, say, Jimi Hendrix, or Usain Bolt and the Tower of London.
Our affinity understanding
To build on our semantic understanding, customer data from social media profiles help to uncover any hidden or overlooked similarities and correlations via what people consume and talk about online.
If left unqualified, every topic in the world would show a huge affinity with the likes of Taylor Swift, Family Guy, Justin Bieber et al, given their popularity on social media.
A normalisation process balances these topics, keeping only genuine links and eliminating incorrect affinities.
Our categorical understanding
Wikipedia is the source of our categorical data because it’s intuitive to humans – and there’s an enormous category structure that already exists.
For example, it’s helpful to bucket Usain Bolt, Carl Lewis, Jackie Joyner-Kersee and Asafa Powell into an ‘Olympic Medal Winners’ or a ‘Track and Field Athletes’ segment. Bolt and Powell would also sub-divide into, say, ‘Famous Jamaicans’, given the country’s small stature.
However, batching Carl Lewis, Jimi Hendrix and Jackie Joyner-Kersee in a ‘Famous Americans’ segment would be too noisy and cumbersome to mean anything so there’s cleaning required to stay relevant to how we, as humans, would naturally categorise topics.
This process is critical for our clients because dependent on the objective they may need to dig deep into niche segmented interests, or stay at a high level.
Visualise & Take Action
These algorithms comprise our Intent Graph and work in real-time to understand what matters to each customer and reveal their interests before taking action.
The Intent Graph is a culmination of a vast machine-learning and data science project designed to provide a human-like understanding of any customer dataset, enabling businesses to cascade scalable customer insight across the entire organisation.
The Insights Dashboard