Work through Patent Documents in a Breeze
... using AI technology for analyzing patent publications to automate highlighting passages describing
scroll down
TPatent publications are typically very verbose and include repetitive text. IP Goggles directs you right to the essential information by highlighting text that can be the key for an argument. IP Goggles saves you time and makes attorneys more efficient in reading patent publications.
With the whole public library of international patent offices at your hand, IP Goggles works in your browser and is integrated with familiar public services like Google Patents or Espacenet offered by the EPO.
IP Goggles uses cloud-based machine learning that can be adapted to specific technology areas and user preferences.
Patent attorneys commonly identify and mark passages in a patent publication describing technical effects and advantages compared to the state of the art. Our tool mimics this practice and highlights relevant passages of a patent text using colors: advantages, conventional technology, problems in the state of the art, boilerplate text. We use the metaphor of virtual goggles, "IP Goggles", that enable viewing patent documents with generated highlighting similar to augmented reality.
IP Goggles is based on a modular software architecture around an extension of a webbrowser (currently Google Chrome): The user obtains a patent text from one of the public patent databases (Google Patents is currently supported). The browser extension analyzes the patent text with the help of cloud-based AI services and highlights the patent text accordingly in the browser – in real-time! The software architecture is flexible to support different front ends for patent databases (e.g. Espacenet) and also different cloud services for analysis.
We use two technologies to identify passages of interest in a patent text: Classification using artificial neural networks. We presently deploy an adapted model of sentiment analysis based on NLP services of Google’s Cloud Platform and linguistic models for semantic matching and search based on a modified version of BERT. The analyses are carried out by cloud services and can be tailored to specific technology areas, e.g. according to IPC classes. Users can feed back their personal markings into the text classifier engines.
IP Goggles is currently not a commercial product.
IP Goggles is developed by members of Betten & Resch for internal use to
suppport attorneys in their daily work and to practice state of the art AI technologies.
If you are interested to learn more about IP Goggles, contact us at ipgoggles@bettenresch.com.
Mauritius Klein
Jan Rietzke
Christoph von Praun
Cris Pana
Christoph Hewel