Sentiment contagion occurs when attitudes toward one topic are influenced by attitudes toward others. Detecting and understanding this phenomenon is essential for analyzing topic evolution and informing social policies. Prior research has developed models to simulate the contagion process through hypothesis testing and has visualized user-topic correlations to aid comprehension. Nevertheless, the vast volume of topics and the complex interrelationships on social media present two key challenges: (1) efficient construction of large-scale sentiment contagion networks, and (2) in-depth explorations of these networks. To address these challenges, we introduce a causality-based framework that efficiently constructs and explains sentiment contagion. We further propose a map-like visualization technique that encodes time using a horizontal axis, enabling efficient visualization of causality-based sentiment flow while maintaining scalability through limitless spatial segmentation. Based on the visualization, we develop CausalMap, a system that supports analysts in tracing sentiment contagion pathways and assessing the influence of different demographic groups. Furthermore, we conduct comprehensive evaluations-including two use cases, a task-based user study, an expert interview, and an algorithm evaluation-to validate the usability and effectiveness of our approach.

The proliferation of large language models (LLMs) has underscored concerns regarding their security vulnerabilities, notably against jailbreak attacks, where adversaries design jailbreak prompts to circumvent safety mechanisms for potential misuse. Addressing these concerns necessitates a comprehensive analysis of jailbreak prompts to evaluate LLMs' defensive capabilities and identify potential weaknesses. However, the complexity of evaluating jailbreak performance and understanding prompt characteristics makes this analysis laborious. We collaborate with domain experts to characterize problems and propose an LLM-assisted framework to streamline the analysis process. It provides automatic jailbreak assessment to facilitate performance evaluation and support analysis of components and keywords in prompts. Based on the framework, we design JailbreakLens, a visual analysis system that enables users to explore the jailbreak performance against the target model, conduct multi-level analysis of prompt characteristics, and refine prompt instances to verify findings. Through a case study, technical evaluations, and expert interviews, we demonstrate our system's effectiveness in helping users evaluate model security and identify model weaknesses.