Reveal hidden insights from your data by connecting the dots
The world we know and live in is deeply connected, though sometimes these connections are not obvious.
If you imagine what you are connected to, you might think of your friends, your family, and places you regularly visit. In business this could be your competitors, suppliers, where you operate, who you sell to and who you want to sell to.
In many cases these data points come from multiple sources, where on their own they tell only part of the story.
Being able to connect all these data points together and articulate the relationships between them gives a richness to the story that enables us to see the whole picture and in doing so - make better decisions.
MODELLING DATA AS A GRAPH
Today, many technologies capture data in table format. This provides a two-dimensional representation of data, which can be enriched by connecting many rows and/or tables together.
Capturing data as a graph extends this capability and enables the rich, relationship-driven structure of the data to be modelled.
The benefits of this approach are:
Seeing the whole picture - data from multiple sources can be modelled in one graph
Deeper understanding of the relationships between data points - directional relationships can be represented
MACHINE LEARNING ON GRAPHS
Once you have the whole picture, the next challenge is to determine what in this data is important to inform your
Machine learning techniques enable us to process the connected data at scale to get:
greater clarity on the contents
reveal hidden connections
prioritise what is important
find with greater speed what we are looking for.
At StellarGraph we employ a number of techniques which you can read about in Machine Learning, though here are some of the most critical to get you started:
Entity resolution - finding out who’s who inside your data.
This is a universal problem across data lakes, or any situation where more than one data set is combined.
Names, addresses, company names and phone numbers can all be written in slightly different ways and are difficult to match. Graph entity resolution uses the rich network data to help uncover the truth, improving and automating the matching accuracy and providing clearer visualisation on how entities are alike.
Predictive modelling - using the relationships of your data to reveal hidden connections and alert areas of interest.
Predictive models can help improve efficiency when resources are limited, by prioritising items and reducing bias in the decision process of what is of greater risk/opportunity.
Graph predictive models can leverage the full graph relationships to enable predictions that are not currently possible with standard techniques.
Data visualisation - enable network data exploration to build trust in machine learning outcomes.
Today, large datasets when visualised are unintelligible to the human eye. Graph machine learning can utilise a number of techniques like summarisation, community detection and pattern detection to cluster data that matters and display it in ways that make sense to users.
The applications for this technology exist wherever high-value connected datasets exist. Think industries like:
Finance | Health | Law Enforcement | Bioinformatics | Cybersecurity.