Browse Prior Art Database

Cognitive Computing Based Smart Prior Art Search Disclosure Number: IPCOM000243453D
Publication Date: 2015-Sep-22
Document File: 3 page(s) / 68K

Publishing Venue

The Prior Art Database


Disclosed is an application of natural language processing to patent prior art review.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 51% of the total text.

Page 01 of 3

Cognitive Computing Based Smart Prior Art Search

The system described in this article is the application of cognitive computing, natural language processing, machine learning, and statistical analysis to shift the labor-intensive process of prior art searching from man to machine.

    Currently, prior art searches are done using fairly rudimentary searches which only process key words and do not take into account phrases and the meaning of the language being used. The searches must be performed by an experienced patent person or team who can give a determination of patentability to a disclosure. The key drawback of people performing prior art searches is time; a person must spend a significant amount of time sifting through the patent disclosure to determine key search terms and then must sort through a massive number of search results from the key terms.

    Speeding up the process of prior art searches has two key advantages: money and time. Every day that is spent processing a disclosure is added time that a competing company could file a patent or defensive submission on the idea. The time people spend searching patents is also expensive; the legal team in charge of making a determination of patentability is composed of highly educated people whose time is wasted performing prior art searches. Some internet sources cite the cost of a prior art search to be between $1,500 and $2,500 per patent. Last year there were almost 550,000 patent applications, meaning the prior art search market is medially valued at $1.3 billion in 2012 to $1.6 billion in 2014.

    A smart prior art search tool utilizing cognitive computing and using the ability to process and comprehend natural language has the ability to output statistical data, with evidence, concerning the patentability of a disclosed idea in the time frame of seconds for a cost calculated in pennies. The bulk of the monotonous labor is shifted to the computer, allowing the patent team to focus their intelligence and time on understanding the value of a patent to the company. The smarter prior art search enables smarter business decisions from the IP team or more rapid and cheaper decisions from the U.S. Patent and Trademark Office.

    This invention consists of the use of a cognitive computer with the ability to process and comprehend natural language to perform smart prior art searches. The computer would ingest a written patent disclosure and use natural language processing to determine both the disclosed idea and its application. The cognitive computer would have previously ingested a patent database and would thus be able to search for similar ideas in similar applications to the disclosure, all due to the abili...