At TalentAdore, we believe in lifelong learning and sharing inspirational and educational content. Thus, this time I have written a comprehensive blog post about AI Matching in recruitment with the help of our CTO, Joni Latvala. The blog post covers the following topics:
- How AI Matching can solve big issues in recruitment
- Practical example of AI Matching in recruitment: TalentAdore’s Recommendation Engine
- How standardized recruitment ontologies support AI Matching
- The benefits of matching candidates with recommendation engine
- How to prevent the AI Matching tools from discriminating against candidates
I hope you’ll enjoy reading the blog 😊💎👇
Recommender systems have become increasingly popular in recent years and have been utilized in a variety of areas, including movies, music, news, books, and products in general. There are also recommender systems for restaurants, financial services, life insurance, romantic partners (online dating), and Twitter pages.
A recommender system or a recommendation system is a subclass of information filtering systems that seeks to predict the rating or preference a user would give to an item.
Recruitment is a very lucrative area for recommender systems. Based on the recommendations, candidates can be matched to the most suitable positions. AI is not, necessarily used in such systems, but it can bring incredible benefits.
AI Matching Can Solve Big Issues in Recruitment
What if we could help job seekers to find the right jobs more easily?
I often hear job seekers say that job seeking is a job itself. It’s hard work. Candidates use their time polishing their application documents, trying to find the right position to apply for, and praying to be noticed in the application process.
“60 percent of job seekers have quit filling out an employment application due to its length or complexity.” (SHRM)
How amazing would it be if a candidate could get automatic recommendations for the most suitable jobs and didn’t need to use her time searching? The best scenario is that the candidate could apply based on the recommendations and would not even need to fill in an application form. That would be an answer to another major pain point in the application process: filling in way too long forms.
What if we could help recruiters to find the right candidates faster and reduce the time-to-hire?
Finding the most suitable employees as easily and quickly as possible is the dream of any organization. Companies struggle to find the right candidates and often end up having too many irrelevant applications. It may look nice as a number in the analytics to have hundreds of applicants, but it doesn’t feel so great when you realize that none of the candidates is the right fit.
“More employers than ever are struggling to fill open jobs — 45% globally say they can’t find the skills they need, up from 40% in 2017 and the highest in over a decade.” (Manpower Group)
While many things can be done to attract more suitable candidates, such as using targeted ads, AI matching can offer one solution to the challenge. Matching tools can be used for automatically recommending the most suitable candidates for the company or a certain position. Candidates can be found and recommended via applications, online platforms, and talent pools or communities.
It is certainly not a foregone conclusion that AI matching will bring benefits. However, when the tools are built the right way and utilize high-quality data, the dreams of both job seekers and recruiters can become a reality.
Recommendation Engine Indicates the Suitability of a Candidate with Colors, Scores and Extracted Skills
TalentAdore has developed an AI-powered Recommendation Engine to match your candidates to positions automatically, giving you recommendations of the most suitable talents. The tool is designed to make your recruitment process more efficient and increase objectivity.
The machine learning algorithm behind the Recommendation Engine uses natural language processing to compare the application document (résumé) and the job description. The similarity measure from 0 to 100 is mapped in the colors. The colors indicate the similarity of an applicant’s background to the job description in question based on the applicant’s résumé, cover letter, and other documentation.
- Green means high similarity
- Yellow means intermediate
- Grey means neutral
When there is insufficient data available—for example, when the résumé is missing, too short, or electronically unreadable—no indication is shown.
The AI matching makes keywords irrelevant by extracting skills from a candidate’s background documentation. The technology also identifies automatically additional skills the candidate has forgotten to mention but which are typically present. In other words, the missing background data is corrected, and you are able to compare apples to apples.
Our AI matching technology has several possible application areas, such as
- Automatically sourcing candidates from your talent communities
- Including a bot-like interactive dialogue on your career pages
- Receiving tips for creating more relevant job ads
Request a demo to find out more, and let’s discuss how we can help you!
AI Matching Is Supported by Standardized Recruitment Ontologies
TalentAdore’s machine learning model has been trained using standardized recruitment ontologies, such as the European Union–based ESCO and O*NET, to produce high-quality and free-of-human-bias results. Also, TalentAdore has accumulated a data set of some hundred thousand of job applicants, many of whom have been reviewed by a recruiter, and that data set is used for validating the model.
ESCO (European Skills/Competences, Qualifications, and Occupations) is a multilingual classification system that covers skills, competencies, qualifications, and occupations. The goal of ESCO is to build a common reference terminology and ontological description to help the European labor market become more effective and integrated and enable worlds of work, education, and training to communicate more effectively with one another.
ESCO is also linked to relevant international classification frameworks, such as the International Standard Classification of Occupations and the European Qualifications Framework. ESCO is a key resource in the Europe 2020 and New Skills Agenda for Europe’s strategic EU projects. Additionally, national employment projects in Finland, such as Työmarkkinatori, utilize ESCO as its backbone for standardizing recruitment, further incentivizing the implementation of ESCO.
The US Department of Labor and Employment and the Training Administration maintain and develop O*NET Online, which contains standardized, occupation-specific descriptions of over 1,000 occupations, covering the entire American economy. O*NET and ESCO are similar in that they drive the same purpose, and O*NET can be nicely aligned with ESCO using ISCO codes.
The Benefits of Matching Candidates with Recommendation Engine
I have previously published a blog post about the benefits of AI in recruitment. The same benefits pretty much fit the AI matching as well, but here is a list of more specific benefits.
1. Remove the Pitfalls of Keyword Searches with Contextual Evaluation
You will no longer overlook candidates. You will get a highlighted recommendation of those candidates who (A) do not have visually impressive CVs and/or (B) did not master the right keywords but who could be a perfect match!
2. Reduce Bias and Improve the Quality of Your Hiring Activity
AI matching can bring an unbiased view to the recruitment process. TalentAdore’s AI cares only about the candidate’s holistic background and skills. Thus, the candidate’s age, gender, or race does not affect the AI score.
3. Attract the Right Candidates and Receive Fewer Irrelevant Applications
You will save time and hire more efficiently with our AI automatically recommending the most suitable positions for your candidates. If the recommendation engine is used before the candidate even applies—for example, as a chatbot—you will get fewer irrelevant applications. Also, your candidates will find the most suitable positions in your organization faster and will be more likely to apply.
How to Prevent the AI Matching Tool from Discriminating against Candidates
We are often asked how can we be sure that the AI is not discriminating against candidates. It is a valid concern, and it deserves an in-depth answer.
Do you know about the Amazon case and what happened with its recruitment tool? To explain briefly, Amazon used artificial intelligence to give job candidates scores ranging from one to five stars. Unfortunately, the tool did not rate candidates for technical jobs in a gender-neutral way; rather, it preferred male candidates.
The problem was that Amazon’s tool was taught with real data that was already biased, reflecting the male dominance across the tech industry. The computer models were trained to score applicants by observing patterns in resumes submitted to the company over a ten-year period.
To prevent AI matching tools from discriminating candidates, it is important to consider these points:
- It is vital to use standardized data (e.g., data from ESCO, O*NET).
- Using the company’s own data for a deep learning model is often not enough. For example, if a company discriminates against women in the job search, a deep learning model will learn to do so as well. At TalentAdore, we do not use company-specific data. We only use real data for validating our model.
- It is important for the user to set certain methods for the profound learning model. One good option is semi-supervised learning, in which a person sanity checks what the model learns.
- AI can help to streamline the recruitment process in many ways, but I do NOT encourage you to reject candidates purely based on AI. At TalentAdore, we use AI to match candidates and write feedback letters. Thus, AI helps the recruiters make the recruitment process more efficient, but the AI does not decide who is selected and who is not.
Thanks for reading! Are you fascinated about AI and can’t wait to learn more? Continue exploring and download our AI in recruitment guide.