While commercial graph databases have been around for more than a decade, it’s the case that they have yet to make much impact in the HR space.
That said, HR tech guru Josh Bersin believes that the technology is a “powerful tool” with much potential in a world that is moving from rigid, hierarchical job descriptions and definitions towards a talent marketplace-based approach.
The focus here is on creating more fluid, flexible roles based on the skills and capabilities required to complete given tasks within a range of projects and assignments.
A key inhibitor to making this vision a reality in the technical sense though is that traditional relational databases are unable to map these complex relationships effectively. Bersin explains:
With relational databases, data is stored in a table with rows and columns. That’s great for things like financial data and where companies are organised into business units and locations as there’s a hierarchical structure to what you’re trying to model. It’s very fast, so you can insert and update data very quickly.
But what it doesn’t do well is map relationships between many things, so when you try to model a complex organization, it’s really hard to do. If you think about a person, for example, you might ask who are they on a team with, who have they worked with in the past and who do they report to now. A graph database can represent that clearly.
But despite the technology’s potential, adoption levels, particularly in the HR world, remain low. As Bersin says:
The technology is pretty mature, but what isn’t is people’s skills and understanding of how to use it. You don’t have research and development groups in HR, so while there are some bigger companies experimenting with it, such as Genpact, a large outsourcing company in India, most companies don’t know the technology exists. So they don’t know they need it because they’ve not seen enough examples of it being used. Those most likely to benefit are organizations like consultancy firms that work on a lot of projects without a linear history of who’s doing what.
Understanding complex organizational relationships at NASA
One organization that is starting to benefit by employing graph databases in the talent space though is the US National Aeronautics and Space Administration (NASA). As part of the space agency’s Future of Work initiative, it created an internal talent marketplace to expand its understanding of, and access to, skills and experience from across the business in order to optimize its staffing of different programs.
But in order to make the process of matching skills with projects more effective, the people analytics team, headed by acting branch chief and senior data scientist David Meza, is currently developing a skills analysis system based on a graph database from supplier Neo4J. He explains:
Part of understanding the Future of Work is understanding talent, so we’re using new technology and algorithms to drive how we can support the workforce and align people across our different missions, especially as we’re trying to get back to the moon and onto mars. This means we need to understand what skills we’ve got and how we can put them to better use. Also when the pandemic hit, it raised questions about how to work in the best way. And these are all graph problems as they’re about relationships between individuals in the world of work, the skills associated with that and the training they need.
The system’s first use case will be to proactively match skills with available openings in the organization’s internal talent marketplace. Meza explains:
With the marketplace, we put in opportunities for short-term projects, and people can enter their profiles to find work they might be interested in doing. The aim is to match people to jobs and jobs to people based on skill sets. But with this application, we’re taking opportunities from the internal marketplace and comparing them with people’s skillsets before sending them a list of suitable recommendations and projects. We can connect several different data sources together to proactively analyse them, which in the past was difficult to do - you had to look through the different opportunities yourself to see if there was a match for what you wanted.
The initiative, which first started in late 2019, is now in the “extended proof of concept” phase, with the people analytics’ team currently working on a front end interface to ensure the application can handle a range of different areas ranging from training assessments and career planning to workforce analytics. The goal is to move the system into production by the end of the year.
While at the moment it is being used by specialist data scientists and analysts to answer questions posed by HR business partners, artificial intelligence and natural language query support mean that over time HR professionals will be able to use it themselves. As the system continues to learn, access will then be opened up more widely to include employees.
Graph databases are game-changers
As for the benefits of using a graph database to underpin the system, Meza believes there are a number:
When you’re handling relationship data from disparate data sources, I don’t think there’s anything better right now. Graph databases make it easier to find information about an employee, the type of work they do, what their skill sets look like across different projects and programmes, whether they’re aligned or whether they could change careers if there are skills gaps. The same kinds of issues exist across all NASA missions, which is about identifying skills sets to ensure the workforce is aligned properly.
But Meza also acknowledges that achieving this aim is a “massive task”, not least because NASA, like many organizations, has realized that:
Up-skilling the workforce is essential, especially when there’s a push for digital transformation and new ways of working. You have to ensure employees are properly trained - and this is one way of understanding if they are. But it’s not just about skills. It’s also about knowledge, task and ability, so we call the database ‘knowledge, skills, ability and technology’.
Another key benefit of graph databases, however, is their ability to deal with relationship-based queries swiftly. For example, a search to find people undertaking specific roles, such as ‘earth scientists’, to showcase the work the agency was doing there at a careers fair, would in the past have taken several hours. Using a graph database, it took less than 15 minutes.
Also valuable though is the ability to see clusters and relationships in previously disparate sources of data. Meza says:
To me, it’s a game-changer in terms of the ability to connect these different sources together to provide information across different domains. It’s much easier to see patterns and relationships using graphs, data science and algorithms to increase usability and capability. While you could do it in a traditional database, it would have taken much longer.
As to what the future holds, a key aim is to increase the ways the system can be used by mapping new data sources into the database. The addition of payroll information, for example, should make it easier to cost projects, while the inclusion of training and adjacent skills information should make it possible to understand the optimum career paths to progress in certain professions.
The fact that innovative organizations, such as NASA, are starting to use graph databases to map and model complex organizational relationships gives some indication of the technology’s potential and may help to kickstart interest in a market that Bersin believes could become signficant.