interaction design

Domain-specific chatbots suffer from ambiguous user intent, contextual fragmentation, and interaction disorganization during multi-turn interactions—such as conditional filtering, multi-option selection, and comparative operations—due to the absence of GUI-like “submit/reset” mechanisms. To address this, this work introduces, for the first time, a form-based Submit/Reset paradigm into conversational systems, explicitly modeling user confirmation behaviors and context-switching actions. Methodologically, we integrate formalized state tracking, fine-grained user action recognition, and chain-of-thought (CoT) reasoning, augmented by prompt engineering to enhance large language models’ capacity for structured dialogue state representation. Experiments in hotel booking and customer management domains demonstrate significant improvements: +28.6% in multi-turn task coherence, +32.1% in user satisfaction, and a reduction of 2.4 turns on average, indicating enhanced operational efficiency.

Contemporary LLM user interfaces predominantly employ linear, rigid interaction paradigms, limiting flexible context management, dynamic dialogue branching, and user-initiated control. To address this, we conducted a design thinking workshop (deRSE25) and propose a human-centered LLM interface paradigm—whiteboard-style, nonlinear, and extensible—emphasizing user agency and contextual adaptability. Our methodology integrates participatory design, iterative user feedback, and visual prototyping tools to systematically explore and empirically validate multiple novel UI concepts. The study yields reusable design principles and an open prototype framework for LLM interaction. More significantly, it advances the development of next-generation LLM interfaces that are human-centric, controllable, and context-aware—thereby enhancing dialogue flexibility and user autonomy in practical deployment. (136 words)

Human-AI conversational interfaces face design challenges stemming from ambiguous user goals, users’ limited awareness of AI capabilities, and the transient, ephemeral nature of interactions. Method: This study proposes an AI-agent-driven CHAI (Conversational Human-AI Interaction) workflow, positioning AI as a collaborative design partner to support real-time prompt generation, user-agent simulation testing, and goal-clarification scaffolding. Developed through four iterative cycles and empirically validated with 10 users, the workflow integrates design probes, thematic analysis, and contextualized prompt engineering. Contribution/Results: We deliver an interpretable prototype suite and introduce the first reusable CHAI agent workflow paradigm under a “research-through-design” framework. Evaluation shows significant improvements in user goal articulation (+42%) and feedback depth. Additionally, we establish a curated, annotated design asset library comprising 12 canonical interaction patterns, enabling high-fidelity early-stage CHAI experience validation.

Large language models (LLMs) exhibit unstable dialogue behavior and poor maintainability in complex business processes. Method: This paper proposes Conversation Routines (CR), a framework that formalizes task-oriented dialogue logic via natural-language specifications, pioneering the integration of structured business workflows directly into LLM prompts—thereby decoupling dialogue design from tool implementation. CR supports modular routine definition and composition, natural-language-driven workflow orchestration, and synergistically combines tool-augmented conversational agents (Tool-Augmented CAS) with prompt engineering. Contribution/Results: Evaluated on two proof-of-concept scenarios—train ticket booking and interactive fault diagnosis—CR enables domain experts to build high-fidelity, high-task-success-rate dialogues without coding. It significantly improves system interpretability, reusability, and cross-role collaboration efficiency.

This study investigates software developers’ functional requirements for large language model (LLM)-driven conversational agents to enhance development efficiency and user experience while supporting differentiated interactions for novices and experts. Through user behavior analysis, semi-structured interviews, and contextualized task evaluations, we systematically identify three core requirements: multi-level experience adaptation, history-aware interaction, and deep code-context understanding—the first empirical characterization of such needs. Based on these findings, we propose design principles for context-aware (CA) conversational agents tailored to software engineering and implement a prototype to validate feasibility. We prioritize three high-impact capabilities: task automation, Git version-control integration, and experience-level–adaptive response generation. Our work delivers the first empirically grounded design guidelines for intelligent programming assistants, addressing critical gaps in human-AI collaborative programming—specifically, requirement modeling and context-adaptive interaction mechanisms.

This study addresses the current lack of a systematic understanding of user interaction mechanisms with large language model–driven computer-use agents and the key design factors influencing their user experience (UX). Through a two-stage approach, the authors construct and empirically validate a UX design space for such agents. First, they synthesize findings from a literature review and expert interviews to develop a taxonomy encompassing dimensions such as user prompting, explainability, and user control. Second, they conduct a Wizard-of-Oz experiment across normal, error, and high-risk scenarios to observe user behaviors, revealing interdependencies among design dimensions and the diversity of user needs. This work presents the first systematically formulated and empirically validated UX design framework for LLM-driven agents, offering developers a structured and actionable foundation for design decisions.

This work proposes a vision-based GUI automation system centered on explicit task planning to address the tendency of existing agents to deviate from user intent in dynamic interfaces and their lack of transparent, intervenable planning mechanisms. By treating task plans as persistent, inspectable, and editable external artifacts, the system adopts a planning-execution decoupled architecture that integrates multimodal visual inputs, screenshot-anchored interventions, and natural language guidance. This design enables real-time monitoring and localized correction of execution trajectories. Experimental results demonstrate that the approach effectively recovers from the majority of automation failures, substantially enhancing the system’s transparency, controllability, and adaptability in complex, evolving graphical user interfaces.

This work addresses the limitations of linear conversation logs generated by existing conversational data analysis systems, which hinder data workers’ ability to retrospect and communicate about nonlinear, iterative analytical processes. To overcome this, the paper proposes a structured dialogue presentation method that introduces probes enabling multi-level navigation, on-demand detail expansion, and context-enhanced summarization—going beyond conventional scrolling and keyword search. By integrating visual recall with sequential and abstraction-based navigation strategies, the approach effectively supports users in recalling, reorienting within, and prioritizing past analytical exchanges. A user study with ten participants demonstrates that the method significantly enhances traceability of analytical reasoning and improves collaborative efficiency, validating its effectiveness in real-world data analysis workflows.

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 work addresses the scarcity of user customization in everyday productivity tools and the lack of flexible, natural language–based mechanisms for adapting system behavior. It proposes embedding generative AI–driven conversational customization into email systems, enabling users to iteratively reshape their inbox structure, interface, and workflows through natural language, thereby transforming static interfaces into malleable data layers. Through a user-centered design probe, the study reveals that users prefer adaptive modifications grounded in existing patterns over creating configurations from scratch. While such customization significantly enhances tool flexibility, it also introduces risks of misconfiguration, necessitating continuous oversight and iterative refinement mechanisms. The findings illuminate practical pathways and design implications for integrating conversational customization into routine productivity applications.