By Rob Hill, CRO at ProFinda
Can a Machine Learning powered Knowledge Graph transform Talent Acquisition?
The true transformation of talent acquisition will only occur when we start to focus the power of Machine Intelligence on understanding and mapping out the knowledge contained within an organisation. This can be achieved by building a knowledge graph of an entire existing workforce (the direct workforce, contingent pool and alumni) and the positions required within an organisation.
Most talent acquisition decisions today are based on 4-10 ‘facts’ relating to skills and potentially competencies. In contrast, by creating a Machine Learning powered Knowledge Graph, organisations are finally able to develop a complete overview of their current workforce, and their roles, with around 60-100 insights relating to their skills, competencies, experiences and aspirations. In doing so, Talent Leaders are able to more accurately map out the missing talents within their organisation.
For many years, all HCM platforms and talent acquisition tools have promised some form of competency framework and matching capability. Yet, every one of these solutions has failed as they are based upon a relational database invented by E.F Codd at IBM from way back in 1970! Plus, they are based upon the faulty assumption that employees and candidates will patiently complete endless fields listing their skills, competencies, aspirations and experiences. Which obviously never happens…
It’s worth us discussing a few technical terms that are becoming more commonplace with talent acquisition.
What is Machine Learning?
Machine Learning is a field of computer science that uses statistical techniques to enable a software algorithm to learn and improve with experience. Within the context of talent management and acquisition a machine learning algorithm can continue to update the skills and experience of employees and candidates, without them having to fill in endless database fields.
What is an ontology and why is it relevant?
An ontology is a formal naming and definition of the types, properties, and inter-relationships of terms. Within the Talent Management sector, the types, properties and interrelationships of all skills, competencies and experiences form a relevant ontology.
At ProFinda we have invested 5+ years and millions of pounds building an ontology of 60,000+ terms used across the corporate world. A simple example is that the ontology will recognise that the skill ‘bitcoin’ is related to ‘cryptocurrency’ and also Ripple and Ethereum.
So what is a Knowledge Graph?
A Knowledge Graph acquires and integrates information from the ontology and, through Machine Learning, applies a reasoner to derive new knowledge for the user. The knowledge graph is the interrelation between work, roles, skills and outcomes and it becomes a self-evolving graph of knowledge.
In their 2012 introduction to the Knowledge Graph, Google announced
“… we’ve been working on an intelligent model — in geek-speak, a ‘graph’ — that understands real-world entities and their relationships to one another: things, not strings.”
Do I need to throw away my existing HR Technology landscape?
No. The point of a Machine-Learning-powered knowledge graph tool is for it to integrate into and between existing platforms to develop and continuously enhance the Knowledge Graph. This graph must grow from the usually very limited 5-10 terms used to describe a person’s skills, competencies and experiences to 40, 60 or even 100 terms.
Where do I find the data? Where do I start?
Whilst most HR platforms have the very limited 5-10 terms used to describe a person’s skills, competencies and experiences, there are often further thin slithers of information scattered across the existing IT landscape (such as between CRM, ERP and platforms). Additionally, information is scattered in external tools like LinkedIn and partially hidden in non-structured sources such as websites, email content or business proposals.
What is needed is a smart technology that is able to integrate all data sources and then ingest the information into the platform. This is the process which moves the needle from the 5-10 terms to the 25-50 terms; before the Machine Learning algorithm can kick in.
There are a few more key concepts to consider when you are planning the transformation of your Talent Acquisition capability.
Knowledge Graphs, by their very nature, have a flexible structure: the ontology is extended and revised as new data arrives. This makes it convenient to store and manage data in a knowledge graph for talent management and acquisition as the terms we use are regularly updated and data growth is critical.
This is even more important when data is arriving from diverse, heterogeneous sources in both structured and unstructured ways. The Knowledge Graph therefore supports a continuously running data pipeline that keeps adding new knowledge to the graph, refining it as new information arrives.
Machine Learning , enhanced by the ontology is then able to recommend new skills, competencies and experiences based upon its growing understanding of the workforce. The platform is able to compare the people with very similar skills, competencies and experiences and make accurate recommendations to other users.
Use Case – Enhancing the Applicant Tracking System (ATS)
Imagine being able to insert a Machine Learning algorithm which uses a well-developed ontology that has produced a Knowledge Graph of 50-100 relevant terms for each member of your existing workforce, plus for roles which are being recruited for, and also the candidate pool. With the engine doing the work – you have a tool that allows:
- Existing employees to be matched to roles which genuinely suit their skills, experience, competencies and aspirations
- A genuine and thorough job description based upon the actual skills and experiences of those currently fulfilling the role
- Real insight into the candidate pool to ensure they are suitable for the role being recruited
The future for talent acquisition is a machine-learning-powered knowledge graph inserted between and around the existing HR technology landscape. By being able to map relationships of data, the platform can add huge value and remove significant cost. The knowledge graph is complemented through Google-esque Natural Language Processing and critically the data is thoroughly detailed and never out of date.
About the author
Rob Hill, CRO at ProFinda is a customer-facing executive leader in digital HR focused on business transformation, client value recognition, customer success and ultimately growing shareholder value. With 20+ years of global Cloud HR technology experience, Rob focusses on delivering genuinely transformational Digital HR solutions which add value by solving real business problems, resulting in amazing customer satisfaction with a genuine ROI.