Meet our Interns!

At Textgain, we like to give creative people a chance to try out innovative and creative ideas that challenge us. We boarded five new interns for the 2018-2019 period, who are working on different projects related to Natural Language Processing. These projects will help tackle hate speech and other important issues of our societies.

 

Elizabeth Cappon joined us from the University of Antwerp and is working with Gijs Van Beek to detect hate speech. Her project is aimed at checking the reliability of human annotations in the task of separating hate speech from mere offensive or neutral speech, in the Dutch language. Through this approach, she will check how this affects the accuracy of automatic detection of racist, anti-LGBT, and sexist rhetoric.

 

Elizabeth Cappon

Similarly, Pierre Voue – another one of our interns – is also working on hate speech detection but in the context of right-wing extremism in English. He will also be researching the ethics involved in the collection and processing of such data. 

 

Andriy Kosar is working on the extraction of actionable insights from open-ended fields of surveys. Kosar will be carrying out his research with the help of Guy De Pauw and will study the use of both supervised and unsupervised techniques to extract these insights.

 

Andriy Kosar

We believe in giving interns the freedom to pursue any project that they deem useful. Our intern Harichandana SNL from KU Leuven is researching on text mining techniques that will help extract information from hotel descriptions and reviews. Using AI technology, the system can then aggregate information for travellers with specific requirements, such as Halal amenities.

 

Harichandana SNL

Wout Vandeberg will also be working with Guy De Pauw to develop classification techniques that will help in the detection of online disinformation, polarization, and political bias for the back-end of the VJF-funded platform kloptditwel.be. This system will employ semantic analysis to detect discussion hotspots on the platform.

 

Wout Vandeberg

Comments are closed.