Patents are increasingly of interest to many industries beyond traditional technology sectors. Many companies, universities, and government entities own portfolios of hundreds or thousands of patents. Extensive amounts of patent data must be assessed when the a jurisdiction’s patent office examines pending patent applications, when judges preside over patent litigation, when firms decide what types of R&D they want to invest in, when firms decide what types of technology they want to license, when patent owners determine whether to sue competitors for patent infringement, and when companies assess how to avoid infringing patents.
There are millions of existing patents documents, and each year 300,000 new patents are granted in the U.S. alone. There are a wide variety of patent tasks that remain primarily manual, requiring patent agents and patent attorneys to filter and analyze huge amounts of data, thereby consuming valuable time, money, and other resources. Artificial intelligence (AI) has tremendous potential to facilitate the processing of patent documents and related patent data. The workshop addresses the use of AI techniques for patents. This includes the use of Machine Learning and Natural Language Processing in patent examination, extracting meaning and information from the text of patents, evaluating patent portfolios, patent litigation analytics, patent citation analysis, and evaluating patent licenses.
However, it is often not straightforward to automate these patent processing tasks, especially those which rely on reading and assessing the meaning of patents. The text of patents are a hybrid legal and technical document with a style distinct from documents such as statutes, judicial opinions, or contracts. Patents frequently contain idiosyncratic terms and language structure dictated by either the relevant scientific field or various legal requirements. This unique format often requires novel AI techniques tailored to the patent domain. The workshop is intended to popularize new methods for addressing longstanding problems in automating patent processing. Our goal is to build a stronger community of researchers exploring these methods, to find synergies among related approaches and alternatives, and to promote opportunities for collaboration.