This article describe the behavior on NSP platforms only
Skills extraction on opportunities
To identify the skills related to a job posting, the process follows four main steps:
- Analyze the job title
An AI model is used to guess the main skills just from the job title, based on examples of typical roles and their related skills. - Analyze the job description
- First, the text is cleaned to keep only the parts that describe the role or required skills.
- Then, the AI identifies the necessary skills from this content, whether they are clearly stated or just implied.
- Match with the client’s skills framework
The identified skills are compared to those already listed in the skills framework to ensure consistency and avoid duplicates. - Select the most relevant skills
From all the skills detected, only the most relevant ones are kept (up to 12), with a focus on technical skills and knowledge.
Skills extraction on employee's experiences
To identify skills from a professional experience, the process includes three main steps:
- Analyze the experience content
- From the title of the experience, the AI guesses the most obvious hard skills linked to that role.
- From the description, the AI infers relevant skills based on what was done, whether explicitly mentioned or implied.
- Match with the client’s skills framework
The detected skills are compared with the company’s existing skills framework to find identical or similar entries and remove duplicates. - Select the most relevant skills
Out of all detected skills (usually 20 to 50), only the most relevant are kept—up to 8 per experience—depending on how detailed the description is and how many skills were already listed. The focus is again on technical skills over soft skills.
Note: The process is similar to the one used for job opportunities, but with two differences: experience descriptions aren’t filtered, and the AI is trained with different examples to better match the nature of personal experience texts.