Search
  • Alexandrea Ridden

Knowledge Graphs: What & Why?

5 days ago Google gave us all a refresher on their Knowledge Graph, but whilst they're busy running the biggest search engine in the world - how can you weaponise this technology for your business?


Making More From What You Have

Many global companies are beginning to, or are already making use of the vast wealth of structured data inside their company.


Data lakes have been built, machine learning models have been productionised, and now many are seeing the benefits. Many are also realising the limitations and challenges of getting more from these investments.


As an industry, we’ve spent a lot of money putting data into cutting-edge, well-architected data structures. Security solved, pipelines integrated, and data scientists extracting real actionable insights; yet we’re still missing something.


A Unified View

Somehow, with all this investment in organising data, many companies are still struggling to answer basic questions:


Who are key opinion leaders in this area?

● What projects are being worked on in our company?

● What does our customers' complete journey look like?

● Was this scientist at that conference?


How can you begin to “know your customer” when they are represented five times across your internal systems?


Often an internal id relates to one customer but one customer can end up with multiple internal ids - as many fraud departments will tell you, it is naïve to think otherwise.


It’s likely that we will all have experienced the effects of a business failing to connect the dots and effectively share data internally. When I was moving from the UK to the Netherlands, I told my bank in good time and they informed me that all my banking mail would be sent to a different UK address (my parents’ home!). A few months later, however, the bank sent a new credit card and pin number to my old UK apartment. Now rented to another person. Upon calling them, I was informed that the credit card system was separate to the debit card system which was why the mistake had occurred.


Building a knowledge graph puts you on the path towards unifying all of your organisation’s data into a single store of query-able collective knowledge.


Creating Structure

Combining Data into One Structure

Using machine learning we have been able to predict the future and create insights we couldn’t have before, but in many cases we are drastically under-utilising the information we already know; such as the information locked away within the PowerPoints, documents and emails on our laptops.


It seems like it’s easier to innovate and make new insights, than it is to utilise our existing knowledge?


How many hours have we lost or spent phoning multiple colleagues to find the answers we as a company have already answered? How many times have people worked on the same projects, solved the same problems or rewritten presentations on the same subjects?


By processing this data, pulling out sentiment, relationships and entities we can start to structure this data and load it into a knowledge graph where it can be queried by anyone in your company. Reducing time and frustration from looking for answers we know exist.


Up To Date

The Path to Wisdom

Your knowledge graph is then able to provide the most complete, accurate and up to the minute results to your queries.


By building more of the data you have into a single information view, any knowledge you gain from that data will be better. Meaning any wisdom you gain will also be better informed and more complete.


Accessible and Logical

A good Knowledge Graph forms the foundations for every other piece of analytics you do; from fraud detection and anti-money laundering detection to marketing or investment research.


You won’t need an in-depth understanding of the complex layout of your organisation’s silo'd data to get the answers you need. You will not have to know which tables to join or wait to speak to the right data expert.


By building a knowledge graph, you’ll have built-in access to existing and incoming knowledge. You’ll be able to simplify your data complexities, making querying your data intuitive and logical for anyone in your organisation.


So...

A unified, living, breathing data store which self-organises into a query-able logical graph, enabling non-technical individuals to utilise a wealth of knowledge. Sounds ideal right?


Now that you know what Knowledge Graphs are (and what they can bring to your business), how do you go about building one? I hope to help demystify this in my next blog.


0 comments