How can machines learn to support recruiters?
HR Tech
December 13, 2022

How can machines learn to support recruiters?

Training an artificial intelligence model, in a nutshell, involves multiple repetitions of a procedure (such as classification) performed with error correction. In some respects, it resembles the human learning process. Artificial intelligence methods use deep neural networks to learn a wide range of tasks including classification and regression to generate results.

Looking at the classifications, for example, a system can distinguish between many species of birds or people with a given job profile. Regression methods allow for predictive trends.

However, to implement the above-mentioned examples on the necessary scale for AI, we have to use neural networks. Such a network consists of multiple layers of neurons connected to each other between the layers from the input value to the output value. Neurons receive information from previous layers and then process it.

In a sense this imitates the biological neural network (your brain), hence its name. By converting input data - such as a picture or document - into numbers and placing them in a mathematical space, you can feed it to the network. Regardless of the problem and output, be it bird classification, text recognition, or stock market prediction, each network must first learn.

The learning process involves repeatedly processing very large amounts of data and correcting parameters so that, for example, it classifies a photo of a bird into the right species or a candidate for the right job.

So, learning can be defined as a repetitive algorithmic process to optimize a neural network, based on a large amount of data.

Now you might be asking yourself: how much data is needed for AI in recruitment management? The answer is ambiguous because it depends a great deal on the specifics of the given industry.

Firstly, in the case of recruitment, we rely on documents - such as a resume or job ad. We can assume that the more data we accumulate over time, the better the system will perform, but the truth is that a well-designed system can achieve great efficiency even on the basis of tens of thousands of files. It is not that much considering that almost all people work at some point in their lives, so the data pool is really big.

Secondly, we can look at the systematics of the labor market itself and note that the division of the labor market into industries can facilitate the normalization of data. It makes the process easier, because certain structures can be directly translated into the system (we don't have to invent and test them).

Next, we should ask ourselves: are we going to try to classify, predict, evaluate, or perhaps all three? Each of these questions increases the amount of training data we need to create a model that reflects reality. So, with thousands of pieces of data, we may end up with a poor model that is prone to under- or overfitting. With tens of thousands of data, we can expect our model to reflect reality well enough to do its job. Of course, the more reliable the data, the higher the probability that the model will improve.

The most effective machine learning methods to boost the recruitment process

The list of machine learning methods that can support the recruitment process is virtually endless and limited only by the innovation of the creator. However, there are currently a few basic and well-known models that work to support recruitment processes. The most popular are:

·        evaluating candidates based on, for example, experience and education

·        ranking candidates against each other

·        searching for desired skills

·        analysis and prediction of the labor market in terms of jobs and salaries.

In fact, the number of applications is infinite and limited only by data and imagination. The first three suggestions mentioned above can significantly speed up a recruiter's work, reducing the amount of time spent on search and analysis. Combining different independent AI tools gives even more advantages - it facilitates the creation of a multitasking system, satisfying many stages of the recruitment management process.



Neural Network (NN) – a system designed to process information whose structure and operating principle are similar to the human nervous system. The most prominent feature of a neural network is its ability to learn from examples and the possibility of automatic generalization of acquired knowledge.

Model – a system of assumptions, concepts and relations between them that makes it possible to describe some aspect of reality in an approximate way.

Unsupervised learning – a type of machine learning that aims to discover patterns in a data set without pre-existing labels and with minimal human intervention. Unsupervised learning assumes that the expected output is not present in the learning data.

Clustering – a concept in data mining and machine learning, derived from the broader concept of model-free classification. Cluster analysis is an unsupervised classification method. It is a method that groups elements that are similar to each other (e.g. in terms of meaning).

Data Normalization – a procedure for pre-processing data to enable cross-comparison and further analysis


Interested? Go to our previous article about using artificial intelligence in recruitment, and learn where does the HR data for machine learning come from?

Grzegorz Reinelt
Machine Learning Engineer
HR tech
artificial intelligence
future of HR