> ## Documentation Index > Fetch the complete documentation index at: https://docs.asendia.ai/llms.txt > Use this file to discover all available pages before exploring further. # Understanding Semantic Search - Talent smart search > Use Asendia’s AI Sourcing Agent to discover top talent with intelligent semantic search ## Introduction to Semantic Search * What is semantic search * How it differs from a keyword or Boolean search * Why modern AI sourcing relies on context, meaning, and relationships between skills, experiences, and roles * The evolution from keyword-matching to contextual understanding *** ## How Semantic Search Works * The AI analyzes candidate data, job descriptions, and contextual signals using large language models (LLMs) * Concepts like vector embeddings and similarity scoring * How the system understands synonyms, related skills, and domain context (e.g., “RN” ≈ “Registered Nurse”) * Continuous learning — improving as more searches and hires occur *** ## Semantic Search vs. Boolean Search: A Comparison | Feature | Boolean / Keyword Search | Semantic (Smart) Search | | :---------------- | :--------------------------------------- | :--------------------------------------------------- | | Logic | Matches exact words or Boolean operators | Understands context, meaning, and intent | | Skill Recognition | Requires manual inclusion of synonyms | Automatically expands related skills | | Example | “(nurse OR RN) AND ‘emergency room’” | “Emergency room nurse” → finds all relevant profiles | | Maintenance | High manual effort | Adaptive and self-improving | | Accuracy | Often broad or irrelevant | High-precision matching based on contextual fit | *** ## Case Study: LinkedIn Recruiter vs. Asendia AI Semantic Search **Example scenario:** * Boolean search in LinkedIn Recruiter:\ `("registered nurse" OR "RN") AND ("emergency room" OR "ER") AND ("Arizona")`\ → Returns candidates with matching keywords, but misses profiles with similar experiences written differently. * Asendia AI Semantic Search:\ Input: “Experienced ER nurse with trauma background in Arizona”\ → Returns candidates with matching experience descriptions even if they never used “ER” or “trauma” explicitly — e.g. “Level 1 hospital nurse,” “critical care nurse,” etc. **Result:**\ Asendia AI delivers *contextually relevant* matches, not just keyword hits. *** ## How Asendia AI Finds the Best-Fit Candidates * Combines **semantic understanding**, **skill clustering**, and **experience mapping** * Evaluates role compatibility using deep embeddings trained on millions of job–candidate pairs * Factors in: * Skill relevance * Seniority level * Industry experience * Location and availability * Produces a **fit score** and **contextual summary** per candidate *** ## Continuous Learning and Feedback Loop * Each recruiter interaction (e.g. shortlist, reject, hire) trains the system * AI learns recruiter preferences, improving future sourcing accuracy * Adaptive intelligence: the sourcing agent becomes more personalized for each client *** ## Integration with Asendia AI Platform * Works natively with Tracker, Bullhorn, or internal ATS systems * Pulls live job data and searches candidate pools automatically * Outputs top matches, ranking scores, and summary insights directly into the ATS *** ## Benefits for Recruiters * Dramatically reduces sourcing time * Uncovers hidden matches not found by keyword search * Improves submission-to-placement ratio * Enables personalized, data-driven outreach *** Screenshot2025 08 26at7 14 34PM Pn