Search for: All records

We propose an unsupervised solution to the Authorship Verification task that utilizes pre-trained deep language models to compute a new metric called DV-Distance. The proposed metric is a measure of the difference between the two authors comparing against pre-trained language models. Our design addresses the problem of non-comparability in authorship verification, frequently encountered in small or cross-domain corpora. To the best of our knowledge, this paper is the first one to introduce a method designed with non-comparability in mind from the ground up, rather than indirectly. It is also one of the first to use Deep Language Models in this setting. The approach is intuitive, and it is easy to understand and interpret through visualization. Experiments on four datasets show our methods matching or surpassing current state-of-the-art and strong baselines in most tasks. 

Accurate information from both sides of the contemporary issues is known to be an `antidote in confirmation bias'. While these types of information help the educators to improve their vital skills including critical thinking and open-mindedness, they are relatively rare and hard to find online. With the well-researched argumentative opinions (arguments) on controversial issues shared by Procon.org in a nonpartisan format, detecting the stance of arguments is a crucial step to automate organizing such resources. We use a universal pretrained language model with weight-dropped LSTM neural network to leverage the context of an argument for stance detection on the proposed dataset. Experimental results show that the dataset is challenging, however, utilizing the pretrained language model fine-tuned on context information yields a general model that beats the competitive baselines. We also provide analysis to find the informative segments of an argument to our stance detection model and investigate the relationship between the sentiment of an argument with its stance.