Advertisement

Natural Language Search As A Brand New Sourcer

Article main image
Feb 19, 2020

Before I joined the world of sourcing, I used Google (because there’s no other search engine, right?) in a completely different way than I do now. I used to “speak” to Google naturally like I was asking another human a question or stating a command, and oftentimes I would get just the answer I was looking for.

But when sourcing, that’s not always a viable option, to say the least.

I’ve begun to grasp the concept of using natural language for search: which means searching for what a candidate writes instead of what I’m looking for.

There are many different ways a candidate might write a phrase or buzzword that you’re looking for. There are singular and plural versions of nouns, past and present tense verbs, and more. I figured out that all of this needs to be taken into consideration if you want a search that brings back an accurate capture of the market.

 

Research

 Instead of searching for:

“Control Engineer” “Closed loop” “control algorithms”

I could search for:

“Control engineer” (“closed loop” OR “Close loop” OR “Close-loop” OR “Closed-loop”) AND (“control algorithms” OR “control algorithm” OR “Controls algorithm” OR “Controls algorithms” OR “Algorithm for control” OR “Algorithms for control” OR “Algorithm for controls” OR “Algorithms for controls”)

…and so forth.  I decided to also dive a bit deeper and do some extra research. I asked myself: What does this technical thing that I don’t understand really mean?

After some digging, I learned a little bit about closed loop control algorithms, and noticed a particular acronym sneaking its way into several control engineer profiles I was squinting at in utter confusion.

“PID”

Huh?

Some research yielded me answers. I came to find out that a PID control is a proportional-integral-derivative controller. Long story short, it is one of the most common types of control loops.

 

Refinement

So I added this into my search:

“Control engineer” (“closed loop” OR “control algorithms” OR “control algorithm” OR “control loop” OR “closed-loop” OR “control-loop” OR “close loop” OR “close-loop” OR “Proportional-integral-derivative” OR “Proportional Integral Derivative” OR “PID”)

More results!!

Of course, I figured out it’s important to take it a step farther. Apparently, not everyone who does the same thing is called the same thing.  Even further, people who are called the same thing, don’t always do the same thing…

Shocker.

I then created a more general search to see what kind of titles were used by the engineers with similar skill sets. I checked out their past titles as well to find out what the engineers I needed actually call themselves. This resulted in a similar rendition of this added Boolean.

(Engineer OR System) (Control OR Simulink) (“closed loop” OR “control algorithms” OR “control algorithm” OR “control loop” OR “closed-loop” OR “control-loop” OR “close loop” OR “close-loop” OR “Proportional-integral-derivative” OR “Proportional Integral Derivative” OR “PID”)

As I scanned through my results, I noticed a lot of irrelevant profiles popping up like the Grinch coming to take Christmas away.

 

Context

After some conversations with a superhero who shall not be named, I realized the importance of context (or really, the lack thereof).  I learned that search terms could be anywhere on profiles I was trying to pull, in any context, no matter how relevant.  Everything from references to managers or co-workers, recommendations given or received by other people, and all sorts of irrelevant places.

“No Joe, I’m not particularly excited to see your profile just because you gave a recommendation to your friend Bob, the control engineer.”

The final string that helped remove many irrelevant profiles was created by using the word “engineer” in quotes (which forces Google to NOT word-stem it into not-so-relevant terms) as well as adding control OR controls OR simulink to the string:

“engineer” (controls OR control OR simulink) (“closed loop” OR “control algorithms” OR “control algorithm” OR “control loop” OR “closed-loop” OR “control-loop” OR “close loop” OR “close-loop” OR “Proportional-integral-derivative” OR “Proportional Integral Derivative” OR “PID”)

You can see the quality of the results when tested with a LinkedIn search in Google:

 

In Conclusion

This past month has been a blast for me to dive into Boolean and learn to search with natural language. There’s a vast amount more to learn than I can imagine at this point, but that’s part of the journey. Discovering the importance of context and searching for what candidates write, not what I’m looking for, is just one of the first steps on that road.