Human-machine collaboration for content analysis in context of Ukranian war

When there is no (sufficient large) training corpus it is usually hard to apply existing text analytics methods. This is even more difficult, when dealing with narratives which are spread across different media and developing over time. Another difficulty is to deal with information on the topic available in different languages and political contexts. In this project, which we will do together with a refugee Ukranian academic, we suggest using to combine methods for semantic text similarity with expert human knowledge in a participatory way. Our training corpus includes news articles containing information on extremism and terrorism.

Output

Conference paper

annotated corpus of news articles

project proposal for the SSF call for Ukraine academic refugees in Sweden

Project Partners:

• Umeå University (UMU), Frank Dignum

Primary Contact: Virginia Dignum, Umeå University