Online discourse surrounding geopolitical crises is volatile and complex. For example, users can often change their opinions, and apply rationales divergently based on the specific scenario under discussion. This paper explores such stance and rationale divergence in social media discussions. We focus on two major ongoing conflicts: the Russia-Ukraine and Israel-Palestine wars. Through this, we identify a set of users who discuss both conflicts, and then label each user’s comments with their stance and associated rationale. Using this unique dataset, we explore how people apply rationales divergently, and evolve their opinions over time. Our research contributes to the CHI community by providing a reusable, rationale-level annotation methodology. Our findings can inform the design of moderation tools, recommender systems, and discussion interfaces. These can be used to surface disagreements, calibrate echo-chamber exposure, and ultimately foster healthier online discourse.

In today’s digital landscape, the prevalence of sensitive online content has made trigger warnings essential. These warnings inform viewers that the content they are about to see contains sensitive artifacts (e.g. violence). This paper studies the use of trigger warnings, exploiting data from two major platforms: Does the Dog Die, a crowdsourcing trigger warnings platform, and IMDb, a media database. We first study how different media types (e.g. films, video games, and TV shows) are labeled with varying trigger warnings and the different co-occurrence patterns among different trigger warnings. We also discover controversy surrounding certain trigger warnings, with inconsistent opinions stated by different people. We further show that different jurisdictions (e.g. USA vs. UK) assign different content ratings (e.g. R-18) for the same media, even when the same trigger warnings are present. Finally, we develop automatic detectors to identify trigger warnings from IMDb text. We achieve F1 scores exceeding 0.7 for all 10 selected trigger warnings.

Online comments within news articles are a key way people share opinions. Discovering insightful comments can, however, be challenging for readers. A solution to this problem is using comment curation, whereby professional editors select the highest quality comments manually --- referred to as ''editor-picks''. This paper studies the growing use of professional editor-curation for user-generated comments. We focus on the New York Times as a case study, using a dataset covering 80k articles. We study the characteristics of editor-pick comments, highlighting how editor criteria vary across news sections (e.g. sports, entertainment). We find that editor-pick comments tend to be longer, more relevant to the article, positive in sentiment, and contain low toxicity. Our analysis further reveals that editors within different news sections exhibit differing criteria when they perform comment selection. Thus, we finally propose a set of models that can automatically identify good candidate editor-picks. Our ultimate goal is to reduce editor and journalistic workload, increasing productivity and the quality of curated comments.

Abstract Deformable medical image registration plays a crucial role in theoretical research and clinical application. Traditional methods suffer from low registration accuracy and efficiency. Recent deep learning-based methods have made significant progresses, especially those weakly supervised by anatomical segmentations. However, the performance still needs further improvement, especially for images with large deformations. This work proposes a novel deformable image registration method based on an attention-guided fusion of multi-scale deformation fields. Specifically, we adopt a separately trained segmentation network to segment the regions of interest to remove the interference from the uninterested areas. Then, we construct a novel dense registration network to predict the deformation fields of multiple scales and combine them for final registration through an attention-weighted field fusion process. The proposed contour loss and image structural similarity index (SSIM) based loss further enhance the model training through regularization. Compared to the state-of-the-art methods on three benchmark datasets, our method has achieved significant performance improvement in terms of the average Dice similarity score (DSC), Hausdorff distance (HD), Average symmetric surface distance (ASSD), and Jacobian coefficient (JAC). For example, the improvements on the SHEN dataset are 0.014, 5.134, 0.559, and 359.936, respectively.