Synthesizing information from multiple documents into structured understanding is inherently iterative, yet current approaches provide limited support. LLM-based systems let users query information but produce structures that users cannot reshape; manual tools like mind maps offer full control but lack intelligent assistance; and commercial tools have begun combining retrieval with user contribution, but not within a unified visual knowledge structure. We present MindTrellis, an interactive visual system that addresses this gap by letting users and AI collaboratively build a knowledge graph combining document-derived and user-contributed knowledge. Users can query the graph to retrieve document-grounded information, and contribute new concepts, relationships, and hierarchical organization to reflect their developing understanding. A multi-agent pipeline coordinates intent disambiguation, knowledge placement, and coherence maintenance across both pathways. In a controlled study where 12 participants created slide decks, MindTrellis outperformed a retrieval-only baseline in knowledge organization and cognitive load, with participants valuing progressive graph expansion and the ability to integrate their own insights.

With the rise of mobile-first consumption, users increasingly engage with data visualizations on mobile devices. However, the vast majority of existing visualizations are originally authored for desktop environments. Due to significant differences in viewport size and interaction paradigms, directly scaling desktop charts often results in illegible text, information loss, and interaction failures. To bridge this gap, we propose an automated framework to adapt desktop-based visualizations for mobile screens. By systematically categorizing the operations involved in the adaptation process, we establish a multi-level design space. This space defines evolution rules spanning from the global topology level, through the reference frame level, down to the visual elements level. Guided by this theoretical framework, we developed Proteus, a large language model–driven multi-agent system that automatically parses the online visualizations, predicts optimal transformation strategies within the design space, and generates equivalent, highly readable visualizations for mobile devices. Case studies and an in-depth user study with 12 participants demonstrate the effectiveness and usability of Proteus.

Creating aesthetically pleasing data visualizations remains challenging for users without design expertise or familiarity with visualization tools. To address this gap, we present DataWink, a system that enables users to create custom visualizations by adapting high-quality examples. Our approach combines large multimodal models (LMMs) to extract data encoding from existing SVG-based visualization examples, featuring an intermediate representation of visualizations that bridges primitive SVG and visualization programs. Users may express adaptation goals to a conversational agent and control the visual appearance through widgets generated on demand. With an interactive interface, users can modify both data mappings and visual design elements while maintaining the original visualization's aesthetic quality. To evaluate DataWink, we conduct a user study (N=12) with replication and free-form exploration tasks. As a result, DataWink is recognized for its learnability and effectiveness in personalized authoring tasks. Our results demonstrate the potential of example-driven approaches for democratizing visualization creation.

Generative text-to-image models, which allow users to create appealing images through a text prompt, have seen a dramatic increase in popularity in recent years. However, most users have a limited understanding of how such models work and often rely on trial and error strategies to achieve satisfactory results. The prompt history contains a wealth of information that could provide users with insights into what has been explored and how the prompt changes impact the output image, yet little research attention has been paid to the visual analysis of such process to support users. We propose the Image Variant Graph, a novel visual representation designed to support comparing prompt-image pairs and exploring the editing history. The Image Variant Graph models prompt differences as edges between corresponding images and presents the distances between images through projection. Based on the graph, we developed the PrompTHis system through co-design with artists. Based on the review and analysis of the prompting history, users can better understand the impact of prompt changes and have a more effective control of image generation. A quantitative user study and qualitative interviews demonstrate that PrompTHis can help users review the prompt history, make sense of the model, and plan their creative process.

We introduce a spatial constraint model to characterize the positioning and interactions in visualizations, thereby facilitating the activation of static visualizations. Our model provides users with the capability to manipulate visualizations through operations such as selection, filtering, navigation, arrangement, and aggregation. Building upon this conceptual framework, we propose a prototype system designed to activate pre-existing visualizations by imbuing them with intelligent interactions. This augmentation is accomplished through the integration of visual objects with forces. The instantiation of our spatial constraint model enables seamless animated transitions between distinct visualization layouts. To demonstrate the efficacy of our approach, we present usage scenarios that involve the activation of visualizations within real-world contexts.

We propose AutoTitle, an interactive visualization title generator satisfying multifarious user requirements. Factors making a good title, namely, the feature importance, coverage, preciseness, general information richness, conciseness, and non-technicality, are summarized based on the feedback from user interviews. Visualization authors need to trade off among these factors to fit specific scenarios, resulting in a wide design space of visualization titles. AutoTitle generates various titles through the process of visualization facts traversing, deep learning-based fact-to-title generation, and quantitative evaluation of the six factors. AutoTitle also provides users with an interactive interface to explore the desired titles by filtering the metrics. We conduct a user study to validate the quality of generated titles as well as the rationality and helpfulness of these metrics.

With the increasing adoption of digitization, more and more historical visualizations created hundreds of years ago are accessible in digital libraries online. It provides a unique opportunity for visualization and history research. Meanwhile, there is no large-scale digital collection dedicated to historical visualizations. The visualizations are scattered in various collections, which hinders retrieval. In this study, we curate the first large-scale dataset dedicated to historical visualizations. Our dataset comprises 13K historical visualization images with corresponding processed metadata from seven digital libraries. In curating the dataset, we propose a workflow to scrape and process heterogeneous metadata. We develop a semi-automatic labeling approach to distinguish visualizations from other artifacts. Our dataset can be accessed with OldVisOnline, a system we have built to browse and label historical visualizations. We discuss our vision of usage scenarios and research opportunities with our dataset, such as textual criticism for historical visualizations. Drawing upon our experience, we summarize recommendations for future efforts to improve our dataset.

We propose Edit-History Vis, a visual analytics system designed to facilitate interactive exploration on Wikipedia edit history at a fine-grained level. The examination of detailed changes in Wikipedia articles is crucial for understanding how authors’ perspectives vary and conflict during the collaborative editing process. However, it is challenging to reveal the details while preserving the heterogeneous attributes of revisions, namely the time, content, and editor. The Edit-History Vis system integrates editor and textual changes of revisions by utilizing a force-directed revision graph that groups revisions based on standpoints. Through this revision graph, users can identify and analyze editing events such as edit wars, vandalism, repair, and normal updates. The effectiveness of the system in analyzing the edit history is validated through a qualitative comparison with prior work and a quantitative rating from a user study.

We propose an automatic pipeline to generate visualization with annotations to answer natural-language questions raised by the public on tabular data. With a pre-trained language representation model, the input natural language questions and table headers are first encoded into vectors. According to these vectors, a multi-task end-to-end deep neural network extracts related data areas and corresponding aggregation type. We present the result with carefully designed visualization and annotations for different attribute types and tasks. We conducted a comparison experiment with state-of-the-art works and the best commercial tools. The results show that our method outperforms those works with higher accuracy and more effective visualization.

In this paper, we propose a novel approach to generate captions for visualization charts automatically. In the proposed method, visual marks and visual channels, together with the associated text information in the original charts, are first extracted and identified with a multilayer perceptron classifier. Meanwhile, data information can also be retrieved by parsing visual marks with extracted mapping relationships. Then a 1-D convolutional residual network is employed to analyze the relationship between visual elements, and recognize significant features of the visualization charts, with both data and visual information as input. In the final step, the full description of the visual charts can be generated through a template-based approach. The generated captions can effectively cover the main visual features of the visual charts and support major feature types in commons charts. We further demonstrate the effectiveness of our approach through several cases.

In this paper, we propose a technique for automatically annotating visualizations according to the textual description. In our approach, visual elements in the target visualization, along with their visual properties, are identified and extracted with a Mask R-CNN model. Meanwhile, the description is parsed to generate visual search requests. Based on the identification results and search requests, each descriptive sentence is displayed beside the described focal areas as annotations. Different sentences are presented in various scenes of the generated animation to promote a vivid step-by-step presentation. With a user-customized style, the animation can guide the audience's attention via proper highlighting such as emphasizing specific features or isolating part of the data. We demonstrate the utility and usability of our method through a user study with use cases.

Interactive visualization and exploration of large spatiotemporal data sets is difficult without carefully-designed data pre-processing and management tools. We propose a novel architecture for spatiotemporal data management. The architecture can dynamically update itself based on user queries. Datasets is stored in a tree-like structure to support memory sharing among cuboids in a logical structure of data cubes. An update mechanism is designed to create or remove cuboids on it, according to the analysis of the user queries, with the consideration of memory size limitation. Data structure is dynamically optimized according to different user queries. During a query process, user queries are recorded to predict the performance increment of the new cuboid. The creation or deletion of a cuboid is determined by performance increment. Experiment results show that our prototype system deliveries good performance towards user queries on different spatiotemporal datasets, which costing small memory size with comparable performance compared with other state-of-the-art algorithms.

In this work, we propose a novel approach of volume visualization without explicit traditional rendering pipeline. In our proposed method, volumetric images can be interactively `reversed' given the volumetric data and a static volume rendered image under the desired rendering effect. Our pipeline enables 3D-navigation on it for exploring the given volumetric data without explicit transfer function. In our approach, deep neural networks, combined usage of Generative Adversarial Networks (GANs) and Convolutional Neural Networks (CNN) are employed to synthesize high-resolution and perceptually authentic images directly, inheriting the desired transfer function and viewing parameter implicitly given by the input images respectively.