proposing new research or business ideas, one must cover substantial
publication or patent material in order to ensure it is indeed new. Often,
researchers look mainly at publications whereas industry tend to look at
patents, though both sources contain such information, and increasingly so as
industries tend to work in a more research based manner.
project aims to look at how we can combine publication and patent information
in order to answer questions like which authors/players are dominant for a
given subject? What are the hubs (networks) for a subject? How are the
publication and patent data interlinked (or not) for a given subject? Which
university-company collaborations are visible through the patents and
A subset of
data used for this study could e.g. be retrieved using scholar.py (on Github)
for retrieval of Google Scholar data and patent data could e.g. be retrieved
using the PatentsView API. Other data sources are also available and can be
specify particular examples of interest, like e.g. 3d printers, additive
manufacturing, or digital twins.
mining, bag of words, recurrent neural nets, transfer learning, network
analysis or machine learning methods can be used to make visualizations and analyses
to answer key questions and combine the data sources.
In collaboration withGeneral Electrics, US
The student should be familiar with machine learning and text mining methods/deep learning, and preferably also network analysis.