CERES Palais, room "Ruhrpott" (4.13)
Picture: John Tuesday / Unsplash
Presented by Sebastian Reimann, subproject C04
In project C04, we have collected large amounts of text from religious online forums that provide an ideal basis for applying machine-learning based methods to retrieve potentially interesting metaphors. Unfortunately, most such efforts to date have concentrated on English texts and have, so far, not been applied to online forum data or other types of computer-mediated communication. Moreover, no system has considered a further distinction of deliberate metaphor, which are particularly relevant for the CRC, and other types of metaphor. Instead, a previous survey by Neidlein et al. (2020) even showed significant weaknesses in detecting novel metaphor, an important sub-type of deliberate metaphor. In my talk, I will present ideas on how existing methods for automatic metaphor detection can be improved, especially when it comes to the detection of novel metaphor and ultimately deliberate metaphor. Furthermore, I seek to show how the relationship between deliberate metaphor and its communicative functions in discourse could be explored, and finally, I would like to demonstrate how the detection of hate speech in online forums could benefit from the automatic detection of deliberate metaphor.