Artificial intelligence in recruitment - AI at the hart of Element ATS candidates database

Continuing the subject of AI in recruitment, I will try to describe the revolution this technology brings in terms of managing candidate databases in applicant tracking systems. The word revolution is not an overstatement here. Artificial intelligence, as the first technology, will allow organizations to immediately, using a minimum amount of human labor, use the full potential of their candidate databases. This has never been possible before.

Candidates are stored in various ways: e-mail boxes, laptops, cloud resources. Sometimes dedicated systems are used to collect and manage CVs, in particular recruitment systems, commonly qualified as ATS (applicant tracking system).

Regardless of the method of storage and size of the candidate database, we are not able to use its full potential without the involvement of human labor. The size of this involvement is directly proportional to the size of the candidate base.

The classic model of searching the candidate database - without artificial intelligence

Imagine that we are working for an engineering company that has a database of 10 thousand candidates. These are various candidates that we have gathered in the course of numerous recruitments for engineering positions. Let’s also assume that we are looking for a candidate who:

  • has four years of experience as a engineering pre-sales;
  • has managed the sales department for at least two years;
  • speaks English at C2 level;
  • lives in Warsaw.

What would the process of finding such a candidate in our database looks like? So far, we have had two possibilities:

  1. The first possibility is that we store a CV in a place or system that does not allow us to search the contents of the documents. In such a case, it is necessary to review each CV personally. It is the most expensive scenario. The cost increases with each new CV uploaded to the candidate database. With a database of 10 thousand candidates, it is the cost of checking each of these 10 thousand CVs by a human.
  2. The second possibility is that we have an ATS system of better class, which allows us to search the content of the CVs, then we can search the candidates database with any sequence of characters, often using logical operators (boolean search). So if we are looking for an engineer, then we will type in, e.g., „engineer” in the search engine, and the system will show us all the CVs that contain the word „engineer”. It is not an ideal solution because we still have to look through each resume and check the context of using the word. The more candidates we find, the more resumes we will have to analyze on our own.

Of course, in both cases, the work related to CV analysis is usually done manually upon receiving the application in ATS. The result of this manual analysis can be a note or a label added to the CV. It is a beneficial solution, although only as good as these notes and labels will be made. Unfortunately, accurate labeling and detailed notes require extra time and, therefore, cost.

Artificial intelligence in recruitment will do it for us

In order to solve our task as effectively as possible, we need to use a system that can not only read the content of the documents (example no. 2 above), but must also parse this content. Parsing the content of a CV is the next stage of the development of applicant tracking systems, that revolutionizes the way we work with candidate databases. What does it mean to parse the content of a CV? I will present it on the example of Element applicant tracking system.

Element recruitment system, a few seconds after the application is submitted by the candidate, reads the CV content. It reads character by character and saves the content of the whole document. Then the process of parsing the content begins. The parsing is based on the AI algorithm making an independent decision about the meaning of particular elements of the CV content. In other words, our ATS first reads the content of the CV and then interprets what is in that CV:

  • first and last name
  • residence
  • employer’s company name
  • job position
  • job description
  • start and end dates of work
  • language knowledge levels
  • skills
  • certificates
  • trainings

Therefore, if the artificial intelligence of Element‚s applicant tracking system encounters the phrase „Jan Kowalski” in the resume, it will decide that this phrase is, in fact, the candidate’s name. If it encounters the phrase „Sales Engineer”, it will probably decide that this is the candidate’s job. I used the word „probably” because AI takes many factors into account when making its decision, such as the context in which the word or phrase is located. After analyzing the entire CV, a structured profile of the candidate is created. Thanks to this structure, our ATS knows exactly what the candidate’s employment history is, at which universities and what faculties he studied, or what language skills he declares.

The artificial intelligence algorithm used in Element ATS is able to parse hundreds of CV documents in fractions of seconds and create structured candidate profiles from them. Thus, a structured database of candidates is created, in which finding a candidate with a specific professional history is a matter of split second.

Three classes of applicant tracking systems (ATS)

Some ATS systems available on the market save CV files without reading their content. When searching for candidates, the user of such a recruitment system can only look through each resume individually and, reading them, determine whether the content of the document contains the desired information. This model is already archaic and very expensive with regard to the cost of work necessary to perform the search for candidates. Unfortunately, we still meet companies that use such solutions. Let us call this type of recruitment system as ATS system class I.

The vast majority of ATS systems not only save CV files but also read their content. By reading the content, the user can enter „sales engineer” in the search engine and find all the candidates who used such a phrase in their CVs. However, the user of such an ATS system does not know in what context the phrase was used and must check it himself by browsing through the content of the document. In such systems the only way to facilitate the search for candidates is to add notes or labels. This will be our ATS system class II .

Finally, a small part of the recruitment systems – including the Element – goes a step further and not only reads the content of CVs, but also parses this content on its own. In other words, such an ATS understands what in the content of a CV is a name, what is a surname, a position, an employer, a start and end date, etc. Such a recruitment system automatically turns the entire database of candidates, regardless of its size, into an ordered collection of information in which you can search for candidates more precisely than ever before, without the need for notes and labels (which, naturally, can still be useful). Such a recruitment system will perform our task in the most efficient way. It will be a ATS system class III.

It is worth noting that from the point of view of the users, class III ATS is also Linkedin. However, Linkedin does not have to use AI algorithms, as its users inform Linkedin what their first name, last name, education, and so on.

Soon, we will return to the topic of artificial intelligence in applicant tracking systems. We will tell you how we created AI in Element.

Related posts:

Hashtags# artificial intelligence candidate database ATS applicant tracking system recruitment system

Maciej Michalewski

Maciej Michalewski

CEO @ Element. Recruitment Automation Software.

Recent posts:

Google Kormo Jobs

Google is testing Kormo Jobs in Asia – a new application which, unlike Google Cards, is strictly dedicated to recruitment.


Future of HR Jobs

In what direction does HR jobs evolve? What HR jobs will exist in a few or a dozen years? These questions are answered by the Harvard Business Review report