We investigate five English NLP benchmark datasets (on the superGLUE leaderboard) and two Swedish datasets for bias, along multiple axes. The datasets are the following: Boolean Question (Boolq), CommitmentBank (CB), Winograd Schema Challenge (WSC), Winogender diagnostic (AXg), Recognising Textual Entailment (RTE), Swedish CB, and SWEDN. Bias can be harmful and it is known to be common in data, which ML models learn from. In order to mitigate bias in data, it is crucial to be able to estimate it objectively. We use bipol, a novel multi-axes bias metric with explainability, to estimate and explain how much bias exists in these datasets. Multilingual, multi-axes bias evaluation is not very common. Hence, we also contribute a new, large Swedish bias-labelled dataset (of 2 million samples), translated from the English version and train the SotA mT5 model on it. In addition, we contribute new multi-axes lexica for bias detection in Swedish. We make the codes, model, and new dataset publicly available.

Detecting harmful content on social media plat-forms is crucial in preventing the negative ef-fects these posts can have on social media users. This paper presents our methodology for tack-ling task 10 from SemEval23, which focuseson detecting and classifying online sexism insocial media posts. We constructed our solu-tion using an ensemble of transformer-basedmodels (that have been fine-tuned; BERTweet,RoBERTa, and DeBERTa). To alleviate the var-ious issues caused by the class imbalance inthe dataset provided and improve the general-ization of our model, our framework employsdata augmentation and semi-supervised learn-ing. Specifically, we use back-translation fordata augmentation in two scenarios: augment-ing the underrepresented class and augment-ing all classes. In this study, we analyze theimpact of these different strategies on the sys-tem’s overall performance and determine whichtechnique is the most effective. Extensive ex-periments demonstrate the efficacy of our ap-proach. For sub-task A, the system achievedan F1-score of 0.8613. The source code to re-produce the proposed solutions is available onGithub

We study the influence of different activation functions in the output layer of pre-trained transformer models for soft and hard label prediction in the learning with disagreement task. In this task, the goal is to quantify the amount of disagreement via predicting soft labels. To predict the soft labels, we use BERT-based preprocessors and encoders and vary the activation function used in the output layer, while keeping other parameters constant. The soft labels are then used for the hard label prediction. The activation functions considered are sigmoid as well as a step-function that is added to the model post-training and a sinusoidal activation function, which is introduced for the first time in this paper.

In this paper, we propose a methodology fortask 10 of SemEval23, focusing on detectingand classifying online sexism in social me-dia posts. The task is tackling a serious is-sue, as detecting harmful content on socialmedia platforms is crucial for mitigating theharm of these posts on users. Our solutionfor this task is based on an ensemble of fine-tuned transformer-based models (BERTweet,RoBERTa, and DeBERTa). To alleviate prob-lems related to class imbalance, and to improvethe generalization capability of our model, wealso experiment with data augmentation andsemi-supervised learning. In particular, fordata augmentation, we use back-translation, ei-ther on all classes, or on the underrepresentedclasses only. We analyze the impact of thesestrategies on the overall performance of thepipeline through extensive experiments. whilefor semi-supervised learning, we found thatwith a substantial amount of unlabelled, in-domain data available, semi-supervised learn-ing can enhance the performance of certainmodels. Our proposed method (for which thesource code is available on Github12) attainsan F 1-score of 0.8613 for sub-taskA, whichranked us 10th in the competition.