Exemplar-based image translation, which aims to transfer the style of an exemplar image to an input semantic image, is challenging and important in many applications. Most current methods build coarse correspondences and overlook extracting faithful style information from the exemplar image, leading to unsatisfactory results with style inconsistent with the exemplar image. In this paper, we propose a novel and efficient style mixture block to extract faithful style information and build reliable correspondences progressively. Specifically, instead of modeling explicit correspondences, we extract faithful style descriptors by considering global information about the exemplar features. Then, we generate coefficients for these style descriptors by modeling the interaction between the exemplar image and the input image, and efficiently compose these descriptors using the coefficients. The efficiency of the style mixture block allows a multi-scale architecture to extract and transform style descriptors at different resolutions, deforming the features of the exemplar image and refining the correspondences progressively. Experimental results on several datasets show that our SMixNet outperforms the current state-of-the-art, and is faster. Code is available for research purposes at https://github.com/Zhangjinso/SMixNet.

Understanding influencers’ perspectives and predicting public sentiment are crucial for event assessment and guidance in computational social systems, enabling more informed decision-making. However, this task is inherently challenging due to the unstructured, context-sensitive, and heterogeneous nature of online communication. To address these challenges, we propose a novel intelligent computational framework, Multi-domain Opinion Leader Agents Emotion Prediction (MOAEP). Our framework comprises three key components: (1) An Automatic Question Generation (AQG) module employing “Who, What, Where, When, Why, and How” (5W1H) questioning to systematically explore topic dimensions; (2) A Multi-domain Opinion Leader Agents (MOA) module that integrates enhanced Large Language Models (LLMs) with Retrieval-Augmented Generation (RAG) to produce domain-specific responses; and (3) An emotion prediction engine that synthesizes agent interactions to forecast collective emotional responses, enabling proactive social computing analysis that surpasses conventional post-event methods. Experimental results demonstrate the framework’s efficacy: the AQG module generates high-fidelity outputs, while the influencer agents maintain consistent performance, achieving an average “Generative Pre-trained Transformer 4” (GPT-4) evaluation score of 6.85 (on a 0–10 scale) across multiple dimensions. In a social media conflict case study, “Russia-Ukraine War”, our framework successfully predicts key influencers’ perspectives and aligns emotional forecasts with observed real-world sentiment trends. These findings underscore the potential of MOAEP to provide actionable insights for decision-making in computational social science.