Represent data such as text as numeric vectors that can be compared by similarity.
<div class="scene"><div class="heading">EMBEDDING SPACE</div><div class="plane"><i class="dot a">고양이</i><i class="dot b">반려묘</i><i class="dot c">자동차</i><i class="dot d">엔진</i><span id="focus" class="focus">의미가 비슷한 점은 가까이</span></div></div>.scene{width:min(94vw,800px);height:min(88vh,326px);padding:clamp(10px,2.7vmin,19px);border:1px solid var(--line);border-radius:13px;background:var(--surface);font:500 clamp(13px,2.4vmin,17px)/1.3 var(--font-sans,sans-serif);position:relative;overflow:hidden}.scene .mono{font-family:ui-monospace,SFMono-Regular,monospace}.scene .muted{color:var(--muted)}.scene .accent{color:var(--accent)}.scene .heading{font-weight:750;color:var(--accent);margin-bottom:clamp(5px,1.6vmin,12px)}.plane{height:76%;position:relative;border:1px solid var(--line);border-radius:8px;background:linear-gradient(90deg,transparent 49.5%,var(--line) 50%,transparent 50.5%),linear-gradient(transparent 49.5%,var(--line) 50%,transparent 50.5%)}.dot{position:absolute;padding:5px 8px;border-radius:20px;background:var(--accent);color:var(--surface);font-style:normal;transition:transform .6s}.a{left:9%;top:12%}.b{left:31%;top:27%}.c{left:64%;top:47%;background:var(--accent-2)}.d{left:78%;top:69%;background:var(--accent-2)}.plane.on .b{transform:translate(-18px,-17px)}.plane.on .d{transform:translate(-18px,-16px)}.focus{position:absolute;left:8px;bottom:5px;color:var(--muted)}const plane=document.querySelector('.plane');function draw(){plane.classList.toggle('on')}draw();setInterval(draw,1450)An embedding model assigns a numeric vector to a sentence or document. Similar inputs often lie near one another in that space, enabling retrieval and classification.
The demo compresses high-dimensional vectors into a two-dimensional illustration. Proximity does not guarantee equivalent meaning, so evaluate retrieval on real data.
Use embeddings to search or group sentences and documents by semantic similarity.