Introduction

Strong Evidence of Use (EOU) is a critical component in the successful monetization of a patent portfolio, as buyers and licensees frequently base their decision on the presence of strong EOU. Producing strong EOU is is hard to do, the common approach has been to hire experts with the relevant technical background and an understanding of claim construction. While the approach works for a small number of patents, for large portfolios the task becomes prohibitively expensive, time consuming and error prone even for experienced patent specialists. Mistakes can result in thousands of dollars in wasted effort and legal costs, and even a single overlooked claim element can severely limit the value of a patent. The alternative for large portfolios relies on using automation. Commercial patent analysis tools automate certain tasks, however none are geared specifically for EOU. Furthermore, the tools are generally expensive, require training and a skilled user. The problem is that these tools typically rely on patent office data and there is simply no information that can be directly applied towards determining the ability to produce EOU. Patents are granted irrespective of whether allowed claims can be shown to have EOU. Some attempts have been made to extract inferences by, for example, using claim word count as an indicator; however, experience has shown that these methods are generally inaccurate. Hence, we developed Pelent – a system optimized for finding patents capable of producing strong EOU.

Our Unique Approach

First, we premise that litigated patents are valuable. While not every valuable patent ends up being litigated, the relationship is sufficiently strong to posit that litigated patents represent a proxy for valuable patents. Second, we premise that litigated patents inherently have strong legal value. Legal value is a measure that a patented invention can be enforced against infringement and is a key component for the monetization of a patent. The legal factors influencing value highly depend on the way claims are written and their ability to stand up to legal scrutiny. Claim language should be explicitly supported within the specification, written to be understood by persons of ordinary skill and – probably most important – be detectable. Simply stated, if one cannot demonstrate evidence of use of an issued claim, there is no case to assert the patent.

EOU is often expressed in the form of a compelling ‘claim chart’ detailing concrete evidence that a patented invention is being used in a product and how the claims are being infringed. Strong EOU is essential for patent assertion, and in order to produce strong EOU, each and every claim element must be detectable. For patent infringement to exist, US patent law requires that each element or limitation of the claim must exist in the accused product or process – a doctrine known as the “all-elements rule.” Patent litigation is notoriously expensive – lawyers and experts spend hundreds and perhaps thousands of hours searching technical papers, and poring over the patent specification and prior art to produce EOU and associated assertion arguments. We therefore conclude that if a given claim element exists in a litigated patent it therefore has a high probability of being detectable.

Consider for example an “IGBT” or Insulated Gate Bipolar Transistor. Many may be unfamiliar with IGBTs, precisely how they are used, and how they can be detected – hence requiring a technical expert. In our approach we assume that if an IGBT claim element exists in a litigated patent claim, it is probably detectable. In practice it is quite common to find the same claim element in several separate litigated patents, which reinforces that the claim element is detectable. Hence, a litigated patent can be used to extract embedded detectable claim elements and the requirement for a technical expert becomes superfluous, as the work has already been done. The concept extends to hundreds and probably thousands of technical and legal experts that have invested time and effort into thousands of litigated patents. In essence litigated patents represent an untapped gold mine for detectable claim elements.

How the system works

Valuable patents are patents that provide economic benefit and can be legally enforced. Many patents have no value either because the inventions are worthless or because the patent is legally handicapped even though it has business value. Conversely, valuable patents have high business value and high legal value. Regardless of an invention’s business value, a patent’s value is significantly diminished without sufficient EOU. In fact, inventions are knowingly being infringed all the time because enforcement cannot be economically justified.

Pelent is an AI rule-based system applying logic and expert knowledge. The system acquires patent data from the USPTO and US patent infringement litigations daily. The algorithm uses a Natural Language Processing (NLP) engine and a rule set optimized for patents. The system analyses each patent and each claim. Claim elements are extracted, indexed and stored in the system’s database. It is important to note that the system does more than simply correlate extracted claim elements with known detectable claim elements. The algorithm determines the importance of the claim elements and weights these accordingly. For example, claim elements that are combined to form a key inventive step have a higher significance. Claim elements are also checked for their occurrence (or not) within the patent specification and if they are identifiable as an understood technical term using a database of known and widely accepted claim constructions (e.g. “a tunneling diffusion region”). The latter also helps identify claim elements that may not exist in the litigated patents database. To analyze any given patent or patent portfolio, the system produces a legal value score for each independent claim. Legal value is expressed as a decimal value on a scale from 0-1 with 1 representing the highest score. The system identifies the best independent claim as well as the number of claim elements for each independent claim. The independent claim with the highest score determines the overall legal value score for the patent. A high legal value score indicates a high probability of producing strong EOU; conversely a low legal value score signals a low probability of producing strong EOU. Since the launch of version 1.0 in 2015, the system’s database has grown considerably, in essence becoming ‘smarter’ with the new technical terms. Currently the database holds over 100,000 litigated patent records. Version 2.0, released in 2019 brought improved performance, reliability, and security. The system runs 24/7 using two high performance cloud-based servers running in tandem and is backed up every 4 hours. For a suitably sized portfolio the analysis typically takes hours versus several days or weeks when done manually. There is no limit on the number of patents analyzed; the system routinely analyzes portfolios with hundreds and thousands of patents.

Summary

After analyzing thousands of patents, producing hundreds of claim charts, and selling and licensing patents we have a pretty good idea what drives patent value. Our raison d’être is to provide a better way to analyze patents by developing an AI system optimized for patents. Results are data driven, accurate and produced quickly, allowing clients to save costs and work smarter. Pelent Checkup Solutions is the only automated system that quantifies the probability of producing strong Evidence of Use and provides an invaluable tool for patent owners, buyers, and licensees.