The Science Behind Lengo and Dice Insights

On Tuesday, February 20th, 2018 our Customer Success Lead for Lengo, Kelsey Beasley, held an interview with Simon Hammond, a Data Science Consultant with our DHI team.  They discussed some of the science behind the data insights that Lengo and Dice offer to our clients, and how it helps our clients set themselves apart as industry experts. The following are excerpts Kelsey put together from their conversation.

 

KB: Hello Simon, thank you for taking some time to answer a few questions with me today. First thing I want to do is help our readers understand a little more about the Data Science team, and the role it plays in the company. In general, how would you describe the team, and your role within it?

SH: The team took root from the various DHI brands, primarily WorkDigital and Dice. I seeded the WorkDigital team a few years ago focusing on extracting insight from the OpenWeb data.

(Note: OpenWeb is a social aggregator tool from Dice. You can learn more about this tool here.)

KB: So we review our databases and find different data points that we can use as metrics when providing insights to our users?

SH: There are a bunch of insights that we pull out of OpenWeb data…One of the most useful is identifying skill sets, i.e. collections of skills that tend to occur together and can be labeled with a general skill name.  Having extracted this knowledge, we can then use this to make searches smarter.

KB: This data also helps us create tools like our Skills Heat Map correct? Can you talk to me about how that tool was created, and how it differs from other seemingly comparable skill maps tools from other sites?

SH: The heat map uses historical Dice data about the number of job postings for each skill, along with the number of profiles containing it. By comparing them, we get an idea of how ‘hot’ it is in terms of supplyvs. demand. Our knowledge of which skill sets a skill can belong to lets us segment the skills, so we can just pull out the relevant terms. Earlier work didn’t segment in this way.

KB: Then this helps provide value and insight to users who may not be experts or familiar with different industry segments personally?

SH: Yes, we can show long-term trends for thousands of skills based on our entire database. More insight than any expert is likely to be able to gather through their own experience, think of it as a satellite view. This also helps us identify which skills are emerging within a certain set and which are potentially being replaced. This is just one strand of the data science work in DHI of course.

KB: What other data visualization tools do you see getting used often, what helps the clients most in your opinion?

SH: Data visualization is a way of interacting with the data to discover things. The standard web search is another way of interacting with the data to discover candidates but it’s tedious, repeatedly chopping and changing search queries with any search terms you can think of is not a game anyone likes to play really. I believe we have the insights that could let us build a more efficient and enjoyable search process.

KB: Is that where automation or AI could potentially start playing a role?

SH: Machine learning is definitely needed as part of the process, but it can’t do everything. It still needs to understand what you want, and it can really only get that through a series of interactions – it’s a dialogue. Intelligence Augmentation is an approach that might be closer to what I’m thinking of.

KB: I haven’t heard that term before, can you explain what Intelligence Augmentation is?

SH: Basically, it’s combining human insight with computational processing in a kind of partnership. A recruiter may not want to look at 10,000 profiles, but they provide feedback on the value of a few particular characteristics which can be used to refine the search. A whimsical metaphor would be the difference between having a robot, and having a robot suit (Iron Man style).

KB: Are there any projects in the works that would be using IA features, or is that something that would be coming further on down the line?

SH: It’s purely R&D at this point, there are no concrete plans to push it into production at the moment.  In the mean time we can use our interactive tools to deliver some of the value in a straightforward way and iterate on it.

KB: Are there any projects you are working on currently that you want to share with our audience?

SH: There are a couple of new projects in the works from our knowledge graph, but they are very under the hood at the moment, still in the experimental stages.  I’m also focused on finding new ways to use data insights to stay ahead of our competition.   The team is exploring more initiatives, but I cannot speak to those myself.

KB: I’ll have to reach out to some others on the team to see what we have cooking! Thank you for your time today, I look forward to see what will be getting released for our clients down the road!