Gilles Jacobs' Blog

I will be presenting new work on economic event extraction at the FNP-FNS workshop @ COLING2020. TL;DR: We created a dataset and system that automatically gets “what, who, when, how much, and why” is happening from economic news text. Turns out this is quite a difficult task, even with using... [Read More]

Successful prevention of cyberbullying depends on the adequate detection of harmful messages. Given the impossibility of human moderation on the Social Web, intelligent systems are required to identify clues of cyberbullying automatically. Much work on cyberbullying detection focuses on detecting abusive language without analyzing the severity of the event nor... [Read More]

We provide a cross-domain evaluation framework for cyberbullying detection research to overcome the limitations of data scarcity in this field. We also show that crowdsourcing data in a simulated lab setting is a valid way of generating data. We have published new research as an extension from the AMiCA project... [Read More]

Historically, setting up deep learning stacks with GPU support has been a hassle. This was because of poorly documented version conflicts between NVIDIA drivers, CUDA, CUDNN and your DL library of choice (PyTorch and TF for me). I remember having spent hours on setting up Tensorflow because its docs didn’t... [Read More]

Job opening for annotator student workers for common-sense investor sentiment in the SENTiVENT project on English news articles. The data annotation is part of the SENTiVENT project on economic event extraction and sentiment analysis. [Read More]