cd /news/artificial-intelligence/similarity-isn-t-relevance-the-hard-… · home topics artificial-intelligence article
[ARTICLE · art-106940] src=dev.to ↗ pub= topic=artificial-intelligence verified=true sentiment=· neutral

Similarity isn't relevance: the hard part of semantic search

Divyakush Punjabi, a full-stack and AI systems engineer, built the GovernAI Research Atlas, a semantic discovery platform that unifies research across papers, repositories, and policy. The platform uses ChromaDB vector search with Sentence-Transformer embeddings for retrieval, but Punjabi emphasizes that a custom relevance score is crucial for ranking, as the closest vector match is not always the most useful result.

read2 min views1 publishedAug 22, 2026

Here's a dirty secret of search: "the closest match" and "the most useful result" are not the same thing. Return the mathematically nearest document and you'll often hand someone something technically related and practically useless. Relevance is a harder problem than similarity — and it's where good search is won or lost.

Getting that right was the core challenge in the GovernAI Research Atlas, a semantic discovery platform I built to unify research across papers, repositories, and policy.

Semantic search gives you a superpower: embed everything into vectors and find items close in meaning, not just wording. But raw nearest-neighbor retrieval has a blind spot. The vector-closest result might be a tangential paper that happens to share vocabulary, while the genuinely useful one sits slightly further out. Distance in embedding space is a proxy for relevance — a good one, but not the whole story.

If you stop at "closest vector," your search is clever and still frustrating. The Atlas runs ChromaDB vector search with Sentence-Transformer embeddings across sources like OpenAlex and GitHub — that's the retrieval layer, the "what's semantically near this query." On top of it sits a custom relevance score that decides what actually surfaces first.

That two-stage shape is the pattern behind every search system worth using:

Retrieval gets the attention; ranking gets the results. Anyone can wire up a vector database and get "semantically similar" documents. Turning that into something that reliably surfaces the most relevant item first — that's the engineering that separates a search demo from a search product.

Building the Atlas taught me that the last mile of search — deciding what deserves the top spot — is where most of the real value hides. The full retrieval-and-ranking architecture is on the project page.

👉 Explore it: www.divyakush.com/projects/governai-research-atlas Divyakush Punjabi — Full-Stack & AI Systems Engineer

🌐 https://www.divyakush.com · 💼 LinkedIn · 💻 GitHub

── more in #artificial-intelligence 4 stories · sorted by recency
── more on @divyakush punjabi 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/similarity-isn-t-rel…] indexed:0 read:2min 2026-08-22 ·