frontend development

Current learning interfaces suffer from a fragmentation among technical robustness, user-centered design, and grounding in educational theory, lacking effective interdisciplinary integration. This work addresses this gap by synthesizing insights from artificial intelligence, human-computer interaction, and the learning sciences to propose a cohesive set of design principles and a research agenda for next-generation learning interfaces centered on human-AI collaboration. By integrating interactive AI technologies, theoretically informed models of learning, and user-centered system design, the study identifies key challenges and articulates a scalable framework that is pedagogically effective, technically reliable, and aligned with learners’ needs. The resulting approach offers both a theoretical foundation and a practical pathway for advancing the next generation of learning technologies.

This study addresses core challenges in human–data interaction in the AI era, including perceptual latency, limited scalability, outdated interaction paradigms, and insufficient reliability and interpretability of generative outputs. It systematically examines the impact of AI technologies on human–data interaction, exploration, and visualization, redefining the role of humans in intelligent analysis. Integrating insights from cognitive science, perceptual theory, and interaction design principles, the work proposes a human-centered interaction framework that synergizes large language models (LLMs), vision-language models (VLMs), and multimodal visualization techniques. Moving beyond traditional evaluation metrics focused primarily on efficiency and scalability, this research advances a cognition-driven analytical paradigm for human–AI collaboration and articulates design principles and future research directions for human-centric intelligent analytics systems in the AI era.

To address the challenge of real-time visualization adaptation to evolving discussions and audience interests in small-scale interactive settings, this paper proposes a context-aware, AI-driven visualization recommendation method. The approach integrates heterogeneous contextual signals—including dataset characteristics, current visual encoding, conversational content, and audience demographics—leveraging natural language understanding and visualization intent inference models to dynamically generate executable recommendations: chart-type switching, data transformation optimization, and multi-source data fusion. Unlike conventional interaction paradigms relying on static command-to-action mappings, our work introduces the first end-to-end, context-driven framework for real-time adaptive visualization generation. Evaluation via a prototype system demonstrates significant improvements in presentation flexibility and response relevance. However, challenges remain in achieving ultra-low-latency responsiveness and resolving ambiguous user intents.

Existing LLM interaction interfaces exhibit significant limitations in multi-model orchestration, workflow customization, collaborative extensibility, and empirical evaluation. To address these challenges, we propose OpenUI—a modular, open-source, locally deployable LLM interface toolkit featuring a novel “dual-path plugin architecture” that decouples model adaptation from functional extension. OpenUI integrates a community-driven plugin marketplace to facilitate distributed development, sharing, and composability. We rigorously evaluate OpenUI through naturalistic social media interaction analysis, structured user studies, and representative application scenarios. Results demonstrate substantial improvements in usability (+37% task completion rate), extensibility (support for 12+ mainstream LLMs and 50+ community plugins), and community engagement (210% growth in active contributors within six months). This work establishes a systematic, scalable, and collaboratively extensible paradigm for open LLM interfaces.

This study addresses the absence of a systematic approach for selecting appropriate human-AI interaction interfaces based on user needs and task complexity. It proposes a three-dimensional classification framework grounded in workflow complexity, AI autonomy, and AI reasoning capability, developed and validated through co-design workshops and longitudinal qualitative user studies. The work establishes, for the first time, a mapping between task complexity and interface modalities, offering actionable guidance for designing context-aware, scalable AI interfaces that support seamless transitions between modalities and enable users to exercise progressive control. The framework also balances operational impact and safety risks in high-autonomy scenarios. By providing product teams with a shared conceptual language, this research enhances the ability to achieve an effective balance between human oversight and AI autonomy in collaborative settings.

This study systematically identifies and organizes sixteen core challenges surrounding visualization in synchronous remote collaboration. Drawing on insights from twenty-nine international experts, it focuses on five collaborative scenarios—exploratory data analysis, ideation, visualization presentation, data-driven decision-making, and real-time monitoring—and, for the first time, categorizes these challenges into four research and development dimensions: technology selection, social factors, AI assistance, and evaluation. Integrating emerging trends in extended reality (XR) and artificial intelligence (AI), the work proposes a structured framework that offers both theoretical grounding and practical guidance for future research and system design in multimodal, visualization-supported collaborative environments.

Current Web-based AI agents predominantly rely on passive textual prompts, limiting their ability to proactively infer user intent and support interactive data analysis and decision-making. To address this gap, this work proposes WebSeek—a hybrid proactive browser extension that enables users to extract data from web pages and construct, transform, and refine tabular, list-based, and visual artifacts on an interactive canvas. WebSeek integrates context-aware AI to provide both proactive guidance and responsive assistance, while prioritizing transparency and user control. This study presents the first integration of proactive and reactive AI guidance within a web browsing environment. An exploratory user study with 15 participants demonstrates the system’s capacity to support diverse analytical strategies and underscores the critical need for controllability and explainability in human-AI collaboration.

This work addresses the high learning cost of complex graphical user interfaces and the limitations of existing assistance methods, which often rely on separate chat windows or require extensive custom development. The authors propose an in-situ assistance paradigm that leverages a browser extension to perform lightweight, reversible interventions on the DOM, enabling dynamic interface restructuring without altering the underlying application logic. They introduce the first DOM-based design space and computational pipeline for in-situ assistance, integrating natural language understanding, UI element localization, and reversible operations to support real-time injection of hints, highlighting of controls, or layout rearrangements on arbitrary web pages. Evaluations on two complex interfaces demonstrate the approach’s efficacy and reliability, with user studies showing significant improvements over the ChatGPTAtlas baseline in both usability and task completion efficiency.

To address the lack of a systematic framework for agent design in mixed-initiative visual analytics, this study proposes the first comprehensive, lifecycle-spanning six-dimensional agent design space—encompassing perception, environment understanding, action capability, communication strategy, role dynamics, and human–agent collaborative reasoning. Grounded in a systematic literature review and cross-case coding analysis of 90 visual analytics systems and 207 agents, we develop an extensible, reusable classification framework. This framework supports both design decisions for new systems and systematic positioning and evaluation of existing ones. It explicitly identifies critical research gaps—including dynamic role switching and formal modeling of collaborative reasoning—thereby providing theoretical foundations and practical guidance for agent-driven visual analytics. (136 words)

This work proposes a novel approach to automatically transform static visualizations into interactive ones through natural language instructions, eliminating the need for user programming and circumventing dependencies on original source code or data. The method leverages a multimodal large language model (MLLM) and introduces three key innovations: a structured “action-modification” interaction design space, a multi-agent intent parser that accurately interprets user requests, and a visualization abstraction transformer that maps semantic intent to concrete interactive behaviors. Through case studies and user interviews, the authors demonstrate that the proposed framework effectively supports diverse interaction scenarios while offering high flexibility and usability, thereby lowering the barrier for non-experts to create interactive visualizations from static figures.