Patents contain technological knowledge and solutions which become economically protected. However, this knowledge and approaches can be applied for free in scientific projects. To make use of the technological solutions of the huge patent corpus to researcher, semantic technologies are employed in this project to combine scientitific and patent knowledge by a patent knowledge graph (PKG). Named Entity Recognition (NER) methods based on supervised learning can be used to identify domain specific entities (i.e. tangible as well as abstract objects). The training data is annotated manually to define the domain specific corpus. With the annotated and supervised domain data, the Patent Knowledge Graph is generated, see figure. This graph serves as the backbone for a search infrastructure (i.e. an API which delivers results on search queries). At the same time, patent and scientific data are connected by entity linking so that search queries (faced by questions) returns answers from both worlds.

The poster includes on overview about the required steps to get the PKG as well as methods and technolgies.