Patent domain ‘ripe for AI algorithms’, says Google

Google has claimed that artificial intelligence (AI) can help the patent industry to add efficiency and insights to business practices while outlining how its own model can help entities find novel patents.

In a blog post shared on Friday, November 20, Google claimed that its bidirectional encoder representation from transformers (BERT) model—a natural language processing algorithm—could be used to determine the novelty of patents and generate classifications.

The BERT model was trained on more than 100 million patent publications from the US and other countries using open-source tooling.

A whitepaper shared alongside the blog described how to use the trained model in a number of ways, including how to “more effectively perform prior art searching to determine the novelty of a patent application, automatically generate classification codes to assist with patent categorisation, and autocomplete”.

In recent years, the industry has begun to use machine-learning (which is a branch of AI) algorithms to add efficiency and insights to business practices, said Google, adding that the BERT model will help the broader patent community in its application of machine-learning.

According to the blogpost, the trained model will help corporate patent departments looking to improve their internal models and tooling with more advanced machine-learning techniques. It will also help patent offices interested in leveraging machine-learning approaches to assist with patent examinations and prior art searching.

The global patent corpus is large, with millions of new patents issued worldwide every year, and complex, with patent applications averaging around 10,000 words and “meticulously wordsmithed”.

Filings are also unique, as they are written in a “highly specialised ‘legalese’ that can be unintelligible” and are highly context dependent, added Google. In addition, patents represent tremendous business value to a number of organisations.

“We believe that the patents domain is ripe for the application of algorithms like BERT due to the technical characteristics of patents as well as their business value,” said Google.

Did you enjoy reading this story?  Sign up to our free daily newsletters and get stories sent like this straight to your inbox