TMvsCovid19


Use of Topic Modelling to improve transparency in Covid19 research
 


In the context of the Corona pandemic, new scientific publications are produced daily, which are often published on preprint servers without going through a peer review process. In order to create transparency about the large number of publications and to make efficient use of them, an intelligent, preferably automated data management is required.     





 

 

The challenge 

Due to its status as a pandemic, the current corona situation creates a high time pressure for scientific publications. The scientifically established "peer-review" process can often not be followed and publications are published on preprint servers without being checked. The evaluation of such un-reviewed papers can be supported by an automated content classification, the so-called topic modelling. Topic Modelling is a machine learning process and can help researchers to identify relevant trends, topics and publications.

 

Our Contribution 

Within the framework of the project, the Fraunhofer ISST complements the previous developments of the "COVID-19 Knowledge Space" project under the direction of the Fraunhofer SCAI, in which a knowledge graph is being developed that links different levels of knowledge with each other (scientific publications, ontologies, text mining results and biomedical databases) and thus provides the basis for heuristics and metho-des of AI. The Fraunhofer ISST complements this graph with the service of a trend analysis and by connecting further data sources.

 

Results  

The extension of the existing knowledge graph by a trend analysis enables the recognition and visualization of trends in different research disciplines. The resulting transparency supports the research community in their respective research projects and offers the possibility to react to trends. By integrating additional data sources, the content density of the Knowledge Graph is further enriched and can be used to support decision-making.

 

Partners     

  • Fraunhofer Institute for Algorithms and Scientific Computing SCAI

 

Funding

  • Supported by the Fraunhofer-Gesellschaft in the framework of the anti-corona programme