Abusive Language Detection using Syntactic Dependency Graphs
Kanika Narang, Chris Brew · 2020
Automated detection of abusive language online has become imperative.Current sequential models (LSTM) do not work well for long and complex sentences while bi-transformer models (BERT) are not computationally efficient for the task.We show that classifiers based on syntactic structure of the text, dependency graphical convolutional networks (DepGCNs) can achieve state-of-the-art performance on abusive language datasets.The overall performance is at par with of strong baselines such as fine-tuned BERT.Further, our GCN-based approach is much more efficient than BERT at inference time making it suitable for real-time detection. *This work was done while the author was at Facebook. 1 https://pewrsr.ch/2XzABRoHow can say they want equality when they see as lesser beings?How can say they want equality when they see as lesser beings?Coref Coref (c) Parse tree of the racist tweet missed by DepGCN.How can say they want equality when they see as lesser beings?How can say they want equality when they see as lesser beings?Coref Coref (d) Coreference resolution of the tweet.