Manual patent searching involved visiting the public search room in the US Patent and Trademark Office and leafing through books of published patents. It was conducted using US patent classification codes. Over time, patent records moved from books to microfilm and CD-ROMs, yet the search methods and tools remained the same (eg, classification codes, document retrieval and reading). However, the probability of missing relevant patents was relatively high as patents were not always filed under the expected classifications.

Patent search evolution

In the 1990s patent documents were digitised and stored in a single database that could be searched by anyone with internet access.

In 2006 Google Patents offered simple keyword searches of patent documents. While the likelihood of overlooking relevant patents decreased, this method was inadequate and lacked precision.

The ‘second age’ of patent searching witnessed the development of multiple keyword searching using Boolean operators. Despite being faster and more precise, this method was also expensive and time consuming.

To deal with the exhaustive patent database, automatic search methods developed which reduced human effort and oversight. For example, patent analytics uses an integrated database of patents and includes information such as:

  • data corrections;
  • transactions (to determine ownership);
  • status updates; and
  • litigation and assertion.

Many IP specialists use patent analytics for their speed and accuracy in directly addressing business questions. They can also provide comprehensive statistical analysis of patent activity in a particular sector and translate and combine large sets of patent data.

Automated search tools are faster, more efficiency, minimise error and provide useful overviews of the existing technological landscape, including any disruptive technologies which could damage a product.

Artificial intelligence (AI)-based patent searches enable users to quickly identify the most relevant technical literature. Keywords are converted into multiple concepts from a database of interrelated concepts. Artificial intelligence is streamlining the search process by matching ideas, not merely keywords. The technology is evolving rapidly, with search tools entering the market which rival human cognition – such tools have the capacity and knowledge to apply logical permutations and combinations to patent data in order generate insight that would be indeterminable to a human researcher.

The evolution of patent searching from manual to AI-based methods is unique. Millions of data sets can now be searched using simpler, faster and more dependable tools.

This article first appeared in IAM. For further information please visit