Semantic search
Findip finds the patents closest in meaning to your question — not documents that merely share your keywords. Even when the wording differs, it surfaces patents that describe the same problem and solution.
How it differs from keyword search
Keyword search
Only finds documents containing your exact words. Search "dendrite" and you miss a patent that wrote "tree-like crystal."
Semantic search
Understands the meaning of a sentence as a vector, so it finds patents about the same technology even in different words. The more you describe the problem and solution in a sentence, the sharper the results.
A real example
Real results from a Korean-language query (as of August 2026). Rankings and scores shift as the index is refreshed.
Query
"A solid electrolyte that suppresses lithium dendrite growth in all-solid-state batteries" (asked in Korean)
The top of the list is held by LG Energy Solution's all-solid-state battery electrolyte membrane patents, clustered together at re-rank scores of 0.95 and above, followed by dendrite-suppression patents from Japanese firms such as Toyota. Around the core idea of "dendrite suppression," it groups documents by meaning even when their wording differs.
Getting sharper results
- Write a single sentence describing the goal and problem, not a bag of keywords
- Capturing "what, why, how" lets the vector engine cluster patents with the same intent
- See the search guide for tips and core concepts for how it works