Inside a professional patent search: keywords, classifications and citations
Keyword searching alone misses relevant patents. Here is how analysts combine classification codes, citation chasing and full-text queries.
By Ragulika IP Editorial · · 7 min read
Patent drafters are famous for avoiding plain language. A "bicycle" becomes a "human-powered two-wheeled conveyance", and a keyword search for bicycle quietly misses it. Professional searches therefore layer several independent techniques.
1. Break the product into features
We start by turning your disclosure into a list of discrete technical features, then decide which combinations are most likely to be claimed. This list drives every later step and becomes the backbone of the claim charts.
2. Build keyword families
For each feature we build synonym families, including the unusual vocabulary patent drafters use, with proximity operators so terms must appear near each other. Truncation catches variants (rotat* for rotate, rotating, rotation).
3. Use classification codes
The Cooperative Patent Classification (CPC) and International Patent Classification (IPC) systems group patents by technology regardless of the words used. Classification searching is the main defence against creative vocabulary, and it works across languages.
4. Chase citations
Once highly relevant patents are found, we follow their backward citations (what they cite) and forward citations (who cites them). Examiners have already done part of the work here.
5. Screen, then analyse
Searches typically return hundreds or thousands of hits. Each is screened by an analyst, the relevant set is de-duplicated by patent family, legal status is checked, and only then does detailed claim analysis begin.
6. Document everything
A search you cannot reproduce is a search you cannot rely on. Every report we deliver includes the full strategy — strings, classes, databases and dates — so your counsel can audit it.
This article is general information, not legal advice. See our disclaimer.

