Margret Plank, Frauke Ziedorn
Gerrit Bruns, Kader Pustu-Iren
Federal Ministry of Education and Research (BMBF)
July 2020 – August 2023
The aim of the project is to develop a novel visual search for patent retrieval based on the automatic recognition of image similarities and text-image references in patent specifications. The innovative, image-based search is based on machine learning methods and aims to increase the findability and visibility of patents and to overcome language and terminology barriers.
The aim of the project is to develop innovative methods for the automatic analysis of visual elements and for the creation of semantic image-text references in patent specifications and to implement them in an integrated patent retrieval system for multimodal searches, taking into account the needs and information search behaviour of patent searchers. Often, the innovation and exploitation potential of a patent can only be recognised by means of an image. This solution also enables domain and patent class independent patent searches and thus new ways of patent exploitation (keyword: cross-domain, cross-industry). The aim is to develop new methods for searching and analysing visual elements in patents and to integrate them into the patent retrieval tool in order to meet the demand for innovations in the field of patent exploitation in consultation with the Leibniz Gemeinschaft. This requires the classification of different image types, recognition of semantic concepts, development of measures for similarity estimation, user-oriented development and design of the tool's user interface.
The TIB is responsible for the following work packages:
- Project Coordination / Project Management
- Preprocessing database
- Basic algorithms and methods of image analysis
- Advanced algorithms and methods of image analysis
- Fraunhofer IAIS
- Leibniz Headquarters
- University Hildesheim (IWIST)