Extracting skills from content to fuel the LinkedIn Skills Graph

LinkedIn uses AI to extract skills from various content sources, such as resumes, job descriptions, and LinkedIn Learning courses, and maps them to the LinkedIn Skills Graph. The AI model segments large pieces of text into meaningful parts and considers where a skill is mentioned and the type of content it is in. For example, a skill mentioned in the qualifications section of a job posting is considered more important. The model uses a trie-based tagger for token-based skill matches and a semantic approach based on language models to understand contextual information. It includes shared and domain-specific modules to handle different types of content. LinkedIn's infrastructure ensures scalability and smooth operation across systems.