At Point Topic, we have a lot of data on the broadband market.
As is true for all companies, different products are built for different purposes and different clients, using different data. Over time, this means that you end up with different sets of definitions and outputs that do not fully align with each other.
That is fine because definitions and formats need to be most suitable to the question a dataset tries to answer.
But the question is, how does one manage to keep these different sets of definitions, outputs and formats aligned?
One simple answer is to use lookup tables to keep different datasets aligned. In SQL databases, foreign keys can be used to have one value reference a value in another table. That works fine, to an extent.
However, this is labour-intensive and, crucially, you cannot use this approach when you’re lining up values as well as the underlying definitions for your values. For example, while we have lots of data on BT and Openreach, and often data for one is representative of the other, lining them up means you need to also be aware how the two are different and that doing so will, at some point, cause complications.
This is where ontologies come in. Assigning definitions to concepts and things, as well as the relationships between them, allows us to classify data and formalise the relationships between different data in a robust, reliable way.
A simple RDF triplet that formalises a conceptual relationship: https://www.researchgate.net/figure/Example-of-RDF-triple_fig2_323401889
In our context, this is especially useful for understand relationships between ISPs, network operators, wholesalers, retailers as well as the real-world, in-the-ground infrastructure they operate and use, and the products they offer to end-users.
Our ontology is useful for representing complex operator-infrastructure relationships
Introducing the Point Topic Ontology
While there have been previous attempts at creating ontologies for the telecoms industry, they have had low visibility, did not make their codebase public or were simply not compatible with our requirements.
Our approach was to build a new ontology, based on earlier academic literature, and building it around our broad catalogue of real-world data and business needs.
Visualising data on build locations allows us to spot trends, such as which operators are sharing the most locations
The ontology is maintained separately from all of our products, with classes and relationships defined and maintained independently, based on best practices. Data gets tagged and then populates this web of definitions, gradually building up a web of data to represent a complete view of the broadband sector as we (and our data) see it.
Networks are tagged by technology
The ontology allows us to build bridges between all of our products. This allows us to have a complete picture of the broadband market in one environment, where we can query data across multiple dimensions to answer questions in a significantly shorter time than before. We are continuing to work to make this picture even more complete and eventually share this ability with our clients.
Growing the Ontology
So far, getting to where we are has been a complex undertaking. Nevertheless, this is a work in progress and there are many dimensions and relationships we have not defined and classified so far.
A goal we wish to achieve very soon is harmonisation with DCAT 3.0, an EU ontology aimed at promoting interoperability for using data across Europe.
Then we want to gradually add our European and finally global data, as well as complement our broadband data with more mobile data. We can’t wait to see those visualisations...
Contributing to Shared Knowledge
On top of working with our own data, we are developing an open version of the ontology based on the academic paper that initially inspired our own ontology. We hope that our ontology will be more broad, updated with more definitions, and show greater compatibility with industry data. We are also actively working on cooperating with academia to drive this forward further.
Common Telecom Ontology: https://github.com/Point-Topic/cto-ontology
Point Topic: https://www.point-topic.com/
Comments