While large language model (LLM)-based user profiling offers significant potential for personalization, most existing approaches rely on empirical heuristics and lack grounding in the psychological mechanism that drive human behavior. In this paper, we introduce TRIPLE (Theory-guided Reasoning for Intent and habIt Profiling with LLMs for pErsonalization), a novel framework that systematically integrates dual-process theory from social psychology into LLM-based user modeling. TRIPLE (1) constructs a habitual behavior profile by identifying repeated patterns over time to model automatic responses; (2) builds an intentional behavior profile by inferring user attitudes, subjective norms and perceived behavioral control based on the Theory of Planned Behavior (TPB); and (3) generates behavioral rationale that reveal the interaction between habitual and intentional processes to predict user behavior in context-specific situations. We evaluate TRIPLE on five personalization tasks from the LaMP benchmark using multiple open-source LLMs. Results show that TRIPLE consistently outperforms existing in-context learning methods, with especially pronounced gains on complex generative tasks such as headline and title generation. Qualitative analyses further demonstrate that the profiles and reasoning paths generated by TRIPLE provide interpretable and psychologically grounded explanations of user behavior. These findings provide strong evidence that incorporating validated behavioral theories into LLM-based personalization enhances both predictive performance and interpretability, paving a way for theory-driven, socio-cognitively informed user modeling.

Beyond hallucinations, Large Language Models (LLMs) can craft deceptive arguments that erode users’ critical thinking, posing a significant yet underexamined societal risk. To address this gap, we develop a taxonomy of eight deceptive persuasion strategies by integrating top-down rhetorical theory with a bottom-up analysis of 3,360 AI-generated messages by four LLM families and examining their effects on user perceptions. Through a large-scale user study (N=602) complemented by a think-aloud protocol, we found that participants were vulnerable to Information Manipulation and Uncertainty Exploitation, especially when a message contradicted their prior beliefs. Vulnerability was significantly higher for participants with low cognitive reflection, low topic knowledge, and low topic involvement. Qualitative analyses further revealed that participants were persuaded by the plausibility of an overall narrative even when they distrusted specific details, interpreting deceptive outputs as logically framed information that broadens perspective. We discuss critical implications of these findings for the design of trustworthy AI systems, adaptive user interfaces, and targeted literacy education.

As personality can be useful in many cases, such as better understanding people’s underlying contexts or providing personalized services, research has long focused on modeling personality from data. However, the development of personality detection models faces challenges due to the inherent latent and relative characteristics of personality, as well as the lack of annotated datasets. To address these challenges, our research focuses on methods that effectively exploit the inherent knowledge of Large Language Models (LLMs). We propose a novel approach that compares contrasting perspectives to better capture the relative nature of personality traits. In this paper, we introduce PADO (Personality-induced multi-Agent framework for Detecting OCEAN of the Big Five personality traits), the first LLM-based multi-agent personality detection framework. PADO employs personality-induced agents to analyze text from multiple perspectives, followed by a comparative judgment process to determine personality trait levels. Our experiments with various LLM models, from GPT-4o to LLaMA3-8B, demonstrate PADO’s effectiveness and generalizability, especially with smaller parameter models. This approach offers a more nuanced, context-aware method for personality detection, potentially improving personalized services and insights into digital behavior. We will release our codes.

Vision transformers use [CLS] token to predict image classes. Their explainability visualization has been studied using relevant information from the [CLS] token or focusing on attention scores during self-attention. However, such visualization is challenging because of the dependence of the interpretability of a vision transformer on skip connections and attention operators, the instability of non-linearities in the learning process, and the limited reflection of self-attention scores on relevance. We argue that the output patch embeddings in a vision transformer preserve the image information of each patch location, which can facilitate the prediction of an image class. In this paper, we propose ICEv2 (ICEv2: $${{{\underline{\varvec{I}}}}}$$ nterpretability, $${{{\underline{\varvec{C}}}}}$$ omprehensiveness, and $${{{\underline{\varvec{E}}}}}$$ xplainability in Vision Transformer), an explainability visualization method that addresses the limitations of ICE (i.e., high dependence of hyperparameters on performance and the inability to preserve the model’s properties) by minimizing the number of training encoder layers, redesigning the MLP layer, and optimizing hyperparameters along with various model size. Overall, ICEv2 shows higher efficiency, performance, robustness, and scalability than ICE. On the ImageNet-Segmentation dataset, ICEv2 outperformed all explainability visualization methods in all cases depending on the model size. On the Pascal VOC dataset, ICEv2 outperformed both self-supervised and supervised methods on Jaccard similarity. In the unsupervised single object discovery, where untrained classes are present in the images, ICEv2 effectively distinguished between foreground and background, showing performance comparable to the previous state-of-the-art. Lastly, ICEv2 can be trained with significantly lower training computational complexity.

While a few public benchmarks have been proposed for training hate speech detection models, the differences in labeling criteria between these benchmarks pose challenges for generalized learning, limiting the applicability of the models.Previous research has presented methods to generalize models through data integration or augmentation, but overcoming the differences in labeling criteria between datasets remains a limitation.To address these challenges, we propose PREDICT, a novel framework that uses the notion of multi-agent for hate speech detection.PREDICT consists of two phases: (1) PRE (Perspectivebased REasoning): Multiple agents are created based on the induced labeling criteria of given datasets, and each agent generates stances and reasons; (2) DICT (Debate using InCongruenT references): Agents representing hate and nonhate stances conduct the debate, and a judge agent classifies hate or non-hate and provides a balanced reason.Experiments on five representative public benchmarks show that PREDICT achieves superior cross-evaluation performance compared to methods that focus on specific labeling criteria or majority voting methods.Furthermore, we validate that PREDICT effectively mediates differences between agents' opinions and appropriately incorporates minority opinions to reach a consensus.

Social anxiety disorder is a psychological condition characterized by excessive nervousness in social situations, such as interpersonal interactions. Exposure therapy has shown benefits in its treatment, and virtual reality (VR) technology has gained much attention for reducing physical and psychological distance and providing additional quantitative evidence from the data generated by standard VR devices (e.g., head-mounted display). Clinical psychology studies have highlighted the importance of mediating variables of social anxiety; however, existing VR-based social anxiety studies have neglected such variables with respect to user experience and data analysis in the context of VR, although these variables could provide insights into the design and use of VR for the treatment of social anxiety. In this study, we focused on two representative mediating variables of social anxiety: (1) the gap between self-presentation motivation and expectancy , and (2) self-focused attention . We used sensor data (e.g., head movement, eye movement, eye gaze, and psychological signals) to investigate the impact of these variables on users' anxiety responses in VR. We developed VR-based Social Anxiety Support Tool (VRST) that reflects the theoretical design elements of effective anxiety provocation. Based on the results of a user study with 30 participants, we confirmed that the mediating variables were associated with social anxiety in the VR environment. We also found that the mediating variables were associated with eye gaze, eye pupil, head movement, and body temperature. Our study results provide researchers, designers, and practitioners with empirical evidence and implications for the use of VR technology and sensor data in the mental health context.

Vision transformers use [CLS] tokens to predict image classes. Their explainability visualization has been studied using relevant information from [CLS] tokens or focusing on attention scores during self-attention. Such visualization, however, is challenging because of the dependence of the structure of a vision transformer on skip connections and attention operators, the instability of non-linearities in the learning process, and the limited reflection of self-attention scores on relevance. We argue that the output vectors for each input patch token in a vision transformer retain the image information of each patch location, which can facilitate the prediction of an image class. In this paper, we propose ICE (Adversarial Normalization: I Can visualize Everything), a novel method that enables a model to directly predict a class for each patch in an image; thus, advancing the effective visualization of the explainability of a vision transformer. Our method distinguishes background from foreground regions by predicting background classes for patches that do not determine image classes. We used the DeiT-S model, the most representative model employed in studies, on the explainability visualization of vision transformers. On the ImageNet-Segmentation dataset, ICE outperformed all explainability visualization methods for four cases depending on the model size. We also conducted quantitative and qualitative analyses on the tasks of weakly-supervised object localization and unsupervised object discovery. On the CUB-200-2011 and PASCALVOC07/12 datasets, ICE achieved comparable performance to the state-of-the-art methods. We incorporated ICE into the encoder of DeiT-S and improved efficiency by 44.01% on the ImageNet dataset over that achieved by the original DeiT-S model. We showed performance on the accuracy and efficiency comparable to EViT, the state-of-the-art pruning model, demonstrating the effectiveness of ICE. The code is available at https://github.com/Hanyang-HCC-Lab/ICE.

One of the significant social problems emerging in modern society is mental illness, and a growing number of people are seeking psychological help. Art therapy is a technique that can alleviate psychological and emotional conflicts through creation. However, the expression of a drawing varies by individuals, and the subjective judgments made by art therapists raise the need to secure an objective assessment. In this paper, we present M2C (Multimodal classification with 2-stage Co-attention), a deep learning model that predicts stress from art therapy psychological test data. M2C employs a co-attention mechanism that combines two modalities-drawings and post-questionnaire answers-to complement the weaknesses of each, which corresponds to therapists' psychometric diagnostic processes. The results of the experiment show that M2C yielded higher performance than other state-of-the-art single- or multi-modal models, demonstrating the effectiveness of the co-attention approach that reflects the diagnosis process.

Recent research on creativity support tools (CST) adopts artificial intelligence (AI) that leverages big data and computational capabilities to facilitate creative work. Our work aims to articulate the role of AI in supporting creativity with a case study of an AI-based CST tool in fashion design based on theoretical groundings. We developed AI models by externalizing three cognitive operations (extending, constraining, and blending) that are associated with divergent and convergent thinking. We present FashionQ, an AI-based CST that has three interactive visualization tools (StyleQ, TrendQ, and MergeQ). Through interviews and a user study with 20 fashion design professionals (10 participants for the interviews and 10 for the user study), we demonstrate the effectiveness of FashionQ on facilitating divergent and convergent thinking and identify opportunities and challenges of incorporating AI in the ideation process. Our findings highlight the role and use of AI in each cognitive operation based on professionals’ expertise and suggest future implications of AI-based CST development.

Fashion style analysis is of the utmost importance for fashion professionals. However, it has an issue of having different style classification criteria that rely heavily on professionals’ subjective experiences with no quantitative criteria. We present FANCY (Fashion Attributes detectioN for Clustering stYle), a human-centered, deep learning-based framework to support fashion professionals’ analytic tasks using a computational method integrated with their insights. We work closely with fashion professionals in the whole study process to reflect their domain knowledge and experience as much as possible. We redefine fashion attributes, demonstrate a strong association with fashion attributes and styles, and develop a deep learning model that detects attributes in a given fashion image and reflects fashion professionals’ insight. Based on attribute-annotated 302,772 runway fashion images, we developed 25 new fashion styles (FANCY dataset 1). We summarize quantitative standards of the fashion style groups and present fashion trends based on time, location, and brand.