Predicting Bad Patents

William Ho

EECS Department
University of California, Berkeley
Technical Report No. UCB/EECS-2017-63
May 11, 2017

http://www2.eecs.berkeley.edu/Pubs/TechRpts/2017/EECS-2017-63.pdf

As a response to the rising frequency and cost of patent disputes, the 2012 America Invents Act established a faster, cheaper method to challenge the validity of patents before the Patent Trial and Appeals Board (PTAB). Despite this advance, patent challenges remain costly and time-consuming. We aim to help reduce the frequency of such disputes by developing an automated predictor of patent quality to help inventors write original, high-quality patents and avoid legal challenges down the road. Using machine learning to analyze thousands of PTAB cases, we built a predictor of patent invalidations and case denials that performs better than the background probability. As the PTAB dispute process becomes more well understood with more case data, we hope to improve the quality of our patent analytics to achieve our goal of reducing patent disputes.

Advisor: Lee Fleming


BibTeX citation:

@mastersthesis{Ho:EECS-2017-63,
    Author = {Ho, William},
    Editor = {Stojanovic, Vladimir},
    Title = {Predicting Bad Patents},
    School = {EECS Department, University of California, Berkeley},
    Year = {2017},
    Month = {May},
    URL = {http://www2.eecs.berkeley.edu/Pubs/TechRpts/2017/EECS-2017-63.html},
    Number = {UCB/EECS-2017-63},
    Abstract = {As a response to the rising frequency and cost of patent disputes, the 2012 America Invents Act established a faster, cheaper method to challenge the validity of patents before the Patent Trial and Appeals Board (PTAB). Despite this advance, patent challenges remain costly and time-consuming. We aim to help reduce the frequency of such disputes by developing an automated predictor of patent quality to help inventors write original, high-quality patents and avoid legal challenges down the road. Using machine learning to analyze thousands of PTAB cases, we built a predictor of patent invalidations and case denials that performs better than the background probability. As the PTAB dispute process becomes more well understood with more case data, we hope to improve the quality of our patent analytics to achieve our goal of reducing patent disputes.}
}

EndNote citation:

%0 Thesis
%A Ho, William
%E Stojanovic, Vladimir
%T Predicting Bad Patents
%I EECS Department, University of California, Berkeley
%D 2017
%8 May 11
%@ UCB/EECS-2017-63
%U http://www2.eecs.berkeley.edu/Pubs/TechRpts/2017/EECS-2017-63.html
%F Ho:EECS-2017-63