Cutting-edge image generation has been praised for producing high-quality images, suggesting a ubiquitous future in a variety of applications. However, initial studies have pointed to the potential for harm due to predictive bias, reflecting and potentially reinforcing cultural stereotypes. In this work, we are the first to investigate how multimodal models handle diverse gender identities. Concretely, we conduct a thorough analysis in which we compare the output of three image generation models for prompts containing cisgender vs. non-cisgender identity terms. Our findings demonstrate that certain non-cisgender identities are consistently (mis)represented as less human, more stereotyped and more sexualised. We complement our experimental analysis with (a) a survey among non-cisgender individuals and (b) a series of interviews, to establish which harms affected individuals anticipate, and how they would like to be represented. We find respondents are particularly concerned about misrepresentation, and the potential to drive harmful behaviours and beliefs. Simple heuristics to limit offensive content are widely rejected, and instead respondents call for community involvement, curated training data and the ability to customise. These improvements could pave the way for a future where change is led by the affected community, and technology is used to positively ”[portray] queerness in ways that we haven’t even thought of”’ rather than reproducing stale, offensive stereotypes.

The Stereotype Content model (SCM) states that we tend to perceive minority groups as cold, incompetent or both. In this paper we adapt existing work to demonstrate that the Stereotype Content model holds for contextualised word embeddings, then use these results to evaluate a fine-tuning process designed to drive a language model away from stereotyped portrayals of minority groups. We find the SCM terms are better able to capture bias than demographic agnostic terms related to pleasantness. Further, we were able to reduce the presence of stereotypes in the model through a simple fine-tuning procedure that required minimal human and computer resources, without harming downstream performance. We present this work as a prototype of a debiasing procedure that aims to remove the need for a priori knowledge of the specifics of bias in the model.

To proactively offer social media users a safe online experience, there is a need for systems that can detect harmful posts and promptly alert platform moderators. In order to guarantee the enforcement of a consistent policy, moderators are provided with detailed guidelines. In contrast, most state-of-the-art models learn what abuse is from labeled examples and as a result base their predictions on spurious cues, such as the presence of group identifiers, which can be unreliable. In this work we introduce the concept of policy-aware abuse detection, abandoning the unrealistic expectation that systems can reliably learn which phenomena constitute abuse from inspecting the data alone. We propose a machine-friendly representation of the policy that moderators wish to enforce, by breaking it down into a collection of intents and slots. We collect and annotate a dataset of 3,535 English posts with such slots, and show how architectures for intent classification and slot filling can be used for abuse detection, while providing a rationale for model decisions.1

The ArgMining 2022 Shared Task is concerned with predicting the validity and novelty of an inference for a given premise and conclusion pair. We propose two feed-forward network based models (KEViN1 and KEViN2), which combine features generated from several pretrained transformers and the WikiData knowledge graph. The transformers are used to predict entailment and semantic similarity, while WikiData is used to provide a semantic measure between concepts in the premise-conclusion pair. Our proposed models show significant improvement over RoBERTa, with KEViN1 outperforming KEViN2 and obtaining second rank on both subtasks (A and B) of the ArgMining 2022 Shared Task.

Wikipedia offers researchers unique insights into the collaboration and communication patterns of a large self-regulating community of editors. The main medium of direct communication between editors of an article is the article’s talk page. However, a talk page file is unstructured and therefore difficult to analyse automatically. A few parsers exist that enable its transformation into a structured data format. However, they are rarely open source, support only a limited subset of the talk page syntax – resulting in the loss of content – and usually support only one export format. Together with this article we offer a very fast, lightweight, open source parser with support for various output formats. In a preliminary evaluation it achieved a high accuracy. The parser uses a grammar-based approach – offering a transparent implementation and easy extensibility.