Question Similarity in Community Question Answering: A Systematic Exploration of Preprocessing Methods and Models
In this talk, I describe a study that systematically combines and compares different approaches to the task of question-question similarity, thereby shedding light on the best practice for preprocessing and vector similarity. In addition, I present an error analysis to gain insight into the differences in performance between the system set-ups and show the results of a real-world evaluation through A/B testing.c.
Florian Kunneman is Assistant Professor at the Social AI group at VU University, with the aim to improve on the conversational competence of chatbots. As a postdoctoral researcher at Radboud University and Tilburg University (2017-2019) he studied sentiment (towards vaccination on Twitter and product aspects in online reviews), text readability prediction and question answering.