Low-dimensional projections of text embeddings support visual analysis of document collections, but their spatial organization may not reflect the relationships an analyst intends to examine. Existing semantic interaction approaches encode semantic intent indirectly through geometric constraints or model updates, limiting interpretability and flexibility. We introduce LLM-augmented semantic steering, which enables analysts to express semantic intent by grouping a small set of example documents within the projection. A large language model externalizes this intent as natural-language representations, selectively extends it to related documents, and incorporates the resulting semantic information into document representations via text augmentation or embedding-level blending, without retraining the underlying models. A case study illustrates how the same corpus can be reorganized under different semantic perspectives, while simulation-based evaluation shows that semantic steering improves global and local alignment with target semantic structures using only minimal interaction. Embedding-level blending further enables continuous and controllable steering of projection layouts. These results position projection spaces as intent-dependent semantic workspaces that can be reshaped through explicit, interpretable, language-mediated interaction.

Interactive spatial layouts empower users to synthesize information and organize findings for sensemaking. While Large Language Models (LLMs) can automate narrative generation from spatial layouts, current collage-based and re-generation methods struggle to support the incremental spatial refinements inherent to the sensemaking process. We identify three critical gaps in existing spatial-textual generation: interaction-revision misalignment, human-LLM intent misalignment, and lack of granular customization. To address these, we introduce Semantic Prompting, a framework for spatial refinement that perceives semantic interactions, reasons about refinement intent, and performs targeted positional revisions. We implemented S-prism to realize this framework. The empirical evaluation demonstrated that S-prism effectively enhanced the precision of interaction-revision refinement. A user study (N = 14) highlighted how participants leveraged S-prism for incremental formalization through interactive steering. Results showed that users valued its efficient, adaptable, and trustworthy support, which effectively strengthens human-LLM intent alignment.

The rise of generative AI has increased attention to voice interfaces. In immersive analytics, we conceptualize this trend as Speech-driven Immersive Analytics. While speech interfaces enable natural interactions, users, especially novices, still face a learning curve in articulating analytic intent and exploring data during the foraging phase. Prior work has primarily addressed these challenges through multimodal interaction or textual disambiguation. We introduce a context-aware Speech-driven Immersive Analytics framework (SIA) as a speech-oriented approach that leverages speech acts to convey actionable intent. This framework (SIA) was designed based on a formative study, a prototype development, three technical studies, and a user study. By extracting speech acts from utterances, SIA infers analytic tasks and embodiment tendencies, then integrates them with spatial, chart, and data context to generate feedforward: previews of potential actions and outcomes. The formative study identified user needs. The technical studies demonstrated that SIA improved the inference quality, enabling context-aware feedforward generation. The user study highlighted that the SIA-based prototype was responsive and intuitive, and feedforward helped users learn during the onboarding phase of data exploration. In particular, the user study identified which feedforward elements participants referenced and how they applied them when expressing intent in immersive analytics. Our key technical findings emphasize that the ensemble model, embedded in the Uncertainty Estimator, improves accuracy and stabilizes task inference. The Projector’s context summary was critical in generating context-aware feedforward. Based on these results, we discuss future research directions for intelligent Speech-driven Immersive Analytics.

Embodiment shapes how users verbally express intent when interacting with data through speech interfaces in immersive analytics. Despite growing interest in Natural Language Interactions (NLIs) for visual analytics in immersive environments, users' speech patterns and their use of embodiment cues in speech remain underexplored. Understanding their interplay is crucial to bridging the gap between users' intent and an immersive analytic system. To address this, we report the results from 15 participants in a user study conducted using the Wizard of Oz method. We performed axial coding on 1,280 speech acts derived from 734 utterances, examining how analysis tasks are carried out with embodiment and linguistic features. Next, we measured Speech Input Uncertainty for each analysis task using the semantic entropy of utterances, estimating how uncertain users' speech inputs appear to an analytic system. Through these analyses, we identified five speech input patterns, showing that users dynamically blend embodied and non-embodied speech acts depending on data analysis tasks, phases, and Embodiment Reliance driven by the counts and types of embodiment cues in each utterance. We then examined how these patterns align with user reflections on factors that challenge speech interaction during the study. Finally, we propose design implications aligned with the five patterns.

Computational notebooks, which integrate code, documentation, tags, and visualizations into a single document, have become increasingly popular for data analysis tasks. With the advent of immersive technologies, these notebooks have evolved into a new paradigm, enabling more interactive and intuitive ways to perform data analysis. An immersive computational notebook, which integrates computational notebooks within an immersive environment, significantly enhances navigation performance with embodied interactions. However, despite recognizing the significance of organizational strategies in the immersive data science process, the organizational strategies for using immersive notebooks remain largely unexplored. In response, our research aims to deepen our understanding of organizations, especially focusing on spatial structures for computational notebooks, and to examine how various execution orders can be visualized in an immersive context. Through an exploratory user study, we found participants preferred organizing notebooks in half-cylindrical structures and engaged significantly more in non-linear analysis. Notably, as the scale of the notebooks increased (i.e., more code cells), users increasingly adopted multiple, concurrent non-linear analytical approaches.

Sensemaking is a complex task that places a heavy cognitive demand on individuals. With the recent surge in data availability, making sense of vast amounts of information has become a significant challenge for many professionals, such as intelligence analysts. Immersive technologies such as mixed reality offer a potential solution by providing virtually unlimited space to organize data. However, the difficulty of processing, filtering relevant information, and synthesizing insights remains. We proposed using eye-tracking data from mixed reality head-worn displays to derive the analyst’s perceived interest in documents and words, and convey that part of the mental model to the analyst. The global interest of the documents is reflected in their color, and their order on the list, while the local interest of the documents is used to generate focused recommendations for a document. To evaluate these recommendation cues, we conducted a user study with two conditions: a gaze-aware system, EyeST, and a “Freestyle” system without gaze-based visual cues. Our findings reveal that the EyeST helped analysts stay on track by reading more essential information while avoiding distractions. However, this came at the cost of reduced focused attention and perceived system performance. The results of our study highlight the need for explainable AI in human-AI collaborative sensemaking to build user trust and encourage the integration of AI outputs into the immersive sensemaking process. Based on our findings, we offer a set of guidelines for designing gaze-driven recommendation cues in an immersive environment.

A growing interest in Immersive Analytics (IA) has led to the extension of computational notebooks (e.g., Jupyter Notebook) into an immersive environment to enhance analytical workflows. However, existing solutions rely on the WIMP (windows, icons, menus, pointer) metaphor, which remains impractical for complex data exploration. Although embodied interaction offers a more intuitive alternative, immersive computational notebooks and embodied data exploration systems are implemented as standalone tools. This separation requires analysts to invest considerable effort to transition from one environment to an entirely different one during analytical workflows. To address this, we introduce ICoN, a prototype that facilitates a seamless transition between computational notebooks and embodied data explorations within a unified, fully immersive environment. Our findings reveal that unification improves transition efficiency and intuitiveness during analytical workflows, highlighting its potential for seamless data analysis.

As immersive technologies evolve, immersive computational notebooks offer new opportunities for interacting with code, data, and outputs. However, scaling these environments remains a challenge, particularly when analysts manually arrange large numbers of cells to maintain both execution logic and visual coherence. To address this, we introduce an embodied composition framework, facilitating organizational processes in the context of immersive computational notebooks. To evaluate the effectiveness of the embodied composition framework, we conducted a controlled user study comparing manual and embodied composition frameworks in an organizational process. The results show that embodied composition frameworks significantly reduced user effort and decreased completion time. However, the design of the triggering mechanism requires further refinement. Our findings highlight the potential of embodied composition frameworks to enhance the scalability of the organizational process in immersive computational notebooks.

Existing research on sensemaking in immersive analytics systems primarily focuses on understanding how users complete analysis within these systems with quantitative and qualitative datasets. However, these user studies mainly concentrate on understanding analysis styles and methodologies from a predominantly novice user study population. While this approach provides excellent initial insights into what users may do within IA systems, it fails to address how professionals may utilize an immersive analytic system for analysis tasks. In our work, we build upon an existing immersive analytics concept - "Immersive Space to Think" to understand how professional user populations differ from novice users in immersive analytic system usage. We conducted a user study with 11 professional intelligence analysts who completed three analysis sessions each. Using our results from this study, we provide deep analysis into how professional users complete sensemaking within immersive analytic systems, compare our findings to previously published findings with a novice user population, and provide insights into how to develop better IA systems to support the professional analyst's strategies within these systems.

The computational notebook serves as a versatile tool for data analysis. However, its conventional user interface falls short of keeping pace with the ever-growing data-related tasks, signaling the need for novel approaches. With the rapid development of interaction techniques and computing environments, there is a growing interest in integrating emerging technologies in data-driven workflows. Virtual reality, in particular, has demonstrated its potential in interactive data visualizations. In this work, we aimed to experiment with adapting computational notebooks into VR and verify the potential benefits VR can bring. We focus on the navigation and comparison aspects as they are primitive components in analysts’ workflow. To further improve comparison, we have designed and implemented a Branching&Merging functionality. We tested computational notebooks on the desktop and in VR, both with and without the added Branching&Merging capability. We found VR significantly facilitated navigation compared to desktop, and the ability to create branches enhanced comparison.

As immersive analytics research becomes more popular, user studies have been aimed at evaluating the strategies and layouts of users’ sensemaking during a single focused analysis task. However, approaches to sensemaking strategies and layouts are likely to change as users become more familiar/proficient with the immersive analytics tool. In our work, we build upon an existing immersive analytics approach-Immersive Space to Think-to understand how schemas and strategies for sensemaking change across multiple analysis tasks. We conducted a user study with 14 participants who completed three different sensemaking tasks during three separate sessions. We found significant differences in the use of space and strategies for sensemaking across these sessions and correlations between participants’ strategies and the quality of their sensemaking. Using these findings, we propose guidelines for effective analysis approaches within immersive analytics systems for document-based sensemaking.

WiFi networks are essential to our daily lives, but their signals are not visible to us. Therefore, it is challenging to evaluate the health of a network or make changes to ensure an optimal configuration. Traditional visualization approaches, such as contour lines, are not intuitive and lead to challenges in the analysis and comprehension of networks. In this paper, we introduce two novel visualizations: Wavelines and Stacked Bars. We then compared these visualizations to the state-of-the-art visualization technique of contour lines. We carried out a user study with 32 participants to validate that our novel visualizations can improve user confidence, accuracy, and completion time for the tasks of router localization, ranking of signal strengths, channel interference, and router coverage. We selected these tasks after extensive discussions with domain experts. We believe that our findings will assist network analysts in visually understanding our increasingly rich signal environments.

Eye gaze patterns vary based on reading purpose and complexity, and can provide insights into a reader’s perception of the content. We hypothesize that during a complex sensemaking task with many text-based documents, we will be able to use eye-tracking data to predict the importance of documents and words, which could be the basis for intelligent suggestions made by the system to an analyst. We introduce a novel eye-gaze metric called ‘GazeScore’ that predicts an analyst’s perception of the relevance of each document and word when they perform a sensemaking task. We conducted a user study to assess the effectiveness of this metric and found strong evidence that documents and words with high GazeScores are perceived as more relevant, while those with low GazeScores were considered less relevant. We explore potential real-time applications of this metric to facilitate immersive sensemaking tasks by offering relevant suggestions.

Existing research on immersive analytics to support the sensemaking process focuses on single-session sensemaking tasks. However, in the wild, sensemaking can take days or months to complete. In order to understand the full benefits of immersive analytic systems, we need to understand how immersive analytic systems provide flexibility for the dynamic nature of the sensemaking process. In our work, we build upon an existing immersive analytic system - Immersive Space to Think, to evaluate how immersive analytic systems can support sensemaking tasks over time. We conducted a user study with eight participants with three separate analysis sessions each. We found significant differences between analysis strategies between sessions one, two, and three, which suggest that immersive space to think can benefit analysts during multiple stages in the sensemaking process.

Deep learning has emerged as a powerful tool for feature-driven labeling of datasets. However, for it to be effective, it requires a large and finely labeled training dataset. Precisely labeling a large training dataset is expensive, time-consuming, and error prone. In this article, we present a visually driven deep-learning approach that starts with a coarsely labeled training dataset and iteratively refines the labeling through intuitive interactions that leverage the latent structures of the dataset. Our approach can be used to (a) alleviate the burden of intensive manual labeling that captures the fine nuances in a high-dimensional dataset by simple visual interactions, (b) replace a complicated (and therefore difficult to design) labeling algorithm by a simpler (but coarse) labeling algorithm supplemented by user interaction to refine the labeling, or (c) use low-dimensional features (such as the RGB colors) for coarse labeling and turn to higher-dimensional latent structures that are progressively revealed by deep learning, for fine labeling. We validate our approach through use cases on three high-dimensional datasets and a user study.

The analysis of vast amounts of network data for monitoring and safeguarding a core pillar of the internet, the root DNS, is an enormous challenge. Understanding the distribution of the queries received by the root DNS, and how those queries change over time, in an intuitive manner is sought. Traditional query analysis is performed packet by packet, lacking global, temporal, and visual coherence, obscuring latent trends and clusters. Our approach leverages the pattern recognition and computational power of deep learning with 2D and 3D rendering techniques for quick and easy interpretation and interaction with vast amount of root DNS network traffic. Working with real-world DNS experts, our visualization reveals several surprising latent clusters of queries, potentially malicious and benign, discovers previously unknown characteristics of a real-world root DNS DDOS attack, and uncovers unforeseen changes in the distribution of queries received over time. These discoveries will provide DNS analysts with a deeper understanding of the nature of the DNS traffic under their charge, which will help them safeguard the root DNS against future attack.