How the matching works

An employment matching problem is a situation where the available work and available jobseekers do not match. This may be caused by regional differences, training issues or changes in the occupational structure.  

At Job Market Finland, this problem is solved by building a good online service and an effective matching system to support it. Matching of supply and demand is achieved when suitable jobs and suitable job seekers seek out and find each other.  

Structural data matching

Structural data matching is based first and foremost on how the persons advertising the job describe the position and how the potential employers describe themselves and their skills. This means that it matters how the job posting or job applicant profile is filled out. At the moment, users are offered suggestions through structural matching.

Structural matching takes into account the person’s work history and skills record, their education and language skills, and the area where they are seeking work. The calculations utilise an enriched ESCO vocabulary (which is supplemented by users’ suggestions), education level data, and location data from Statistics Finland. For the person’s work history, the most recent work experiences are emphasised, and minimum competence levels are also taken into account if these have been included in the job posting.

In the matching of structural information, matching points are accumulated by having relevant vocational history and skills. Language skills, education and location, on the other hand, impact the calculated score by deducting points if they do not match.

In other words, the top suggestions will be comprised of job posting or job applicant profiles with suitable training requirements which are located in your desired area and include many matching professions and skills.  

ESCO competences as the basis of matching

The professions and skills displayed at Job Market Finland are based on the ESCO vocabulary. ESCO is an EU-wide vocabulary maintained by the European Commission.

You can find the ESCO glossary and general additional information on the European Commission's website. Professions, competences, and their explanations are presented there in English.

Matching takes place in two different directions

A model based on natural language processing (NLP) was introduced in the spring 2021 alongside the current structural matching methods.

NLP matching is based on written texts and is carried out in three different languages, and machine translation is used to create training data for the different languages.

NLP matching utilises job posting data from Job Market Finland. Meaningful words are extracted from job postings using word vectors and neural networks.

When calculating matching points, word vectors are used to understand which words are related to each other and which job postings are relevant to each job search profile. The score therefore indicates the level of relevance.

For the user, these two methods of matching operate together seamlessly.

Artificial intelligence provides assistance in language recognition

Artificial intelligence can help with the matching – for example, there is a ‘Skills Suggester’ feature, which assists the user in adding the right skills to their job applicant profile or job posting. This makes it easier for you to find the right words to describe your skills.  

Natural language is also utilised by the Skills Suggester. The Skills Suggester is a feature that suggests occupations and skills to you based on natural language, which in this case means written text. 

For example, you can copy the text of a Linkedin presentation or job posting to Job Market Finland to get suggestions for suitable job titles.  

The Skills Suggester makes use of text materials from the ESCO ontology and Ammattinetti. Using machine translation, the methods can be trained for all three languages: Finnish, Swedish and English. TF-IDF document vectors are used to formulate the suggestions.

If you are interested in the technical aspects of the matching system, Developer Heikki Niittylä explains in more detail in his blogs how features such as the Skills Suggester have been built. More explanations for the use of artificial intelligence can be found, for example, on Minna Vänskä’s blog.