Abstract Interpersonal communication in data science can yield sought‐after insights, but presentation environments are often not conducive for live analysis, forcing the process to move offline. Through a formative survey with 16 participants, we identified both technical (e.g., complexity of tools) and psychological (e.g., pressure of programming during presentation) factors constraining live data analysis. To enable live analysis, we present SlideSAVR, a data‐driven presentation assistant that leverages sketching and voice inputs in live discussion to support collaborative data analysis during presentations. Powered by an agentic framework that flexibly defines augmentation rules, updates slide content dynamically to match the live context, and automates backend computations, SlideSAVR enables fluid audience‐presenter interaction and reduces the need for offline reanalysis and follow‐up communication. We demonstrate SlideSAVR's ability to support a range of tasks through nine representative use cases. We further evaluate the system's accuracy and computation time across different settings, showing that SlideSAVR can reliably perform diverse tasks when provided with both sketch and voice inputs.

—Following the rapid emergence of large language models, Multi-Agent Systems (MASs) became a promising approach for accomplishing complex tasks. In MASs, multiple autonomous agents with predetermined roles collaborate by dividing responsibilities. However, MAS developers often struggle to understand and diagnose agents’ behavior from thousands of inter-agent messages across multiple complex conversations. To identify key requirements and challenges related to evaluating, debugging, and managing MASs, we conducted a formative study with six MAS developers. We then introduce ConvoMap, a prototype that addresses a key challenge of MAS development-understanding agents’ behaviors across multiple conversations. ConvoMap integrates automated qualitative coding to enable multi-level inspection of agents’ behavior. ConvoMap can then visualize hundreds of MAS conversations by representing messages as points on a 2D map that encode their semantic meanings and interactions between agents. To better support navigation and deeper analysis, ConvoMap provides topic overviews and highlights relevant text segments. A comparison study showed that ConvoMap helped to understand agents’ behavior more accurately than the baseline.

Analytic provenance can be visually encoded to help users track their ongoing analysis trajectories, recall past interactions, and inform new analytic directions. Despite its significance, provenance is often hardwired into analytics systems, affording limited user control and opportunities for self-reflection. We thus propose modeling provenance as an attribute that is available to users during analysis. We demonstrate this concept by modeling two provenance attributes that track the recency and frequency of user interactions with data. We integrate these attributes into a visual data analysis system prototype, ProvenanceLens, wherein users can visualize their interaction recency and frequency by mapping them to encoding channels (e.g., color, size) or applying data transformations (e.g., filter, sort). Using ProvenanceLens as a design probe, we conduct an exploratory study with sixteen users to investigate how these provenance-tracking affordances are utilized for both decision-making and self-reflection. We find that users can accurately and confidently answer questions about their analysis, and we show that mismatches between the user's mental model and the provenance encodings can be surprising, thereby prompting useful self-reflection. We also report on the user strategies surrounding these affordances, and reflect on their intuitiveness and effectiveness in representing provenance.

Large Language Models (LLMs) have been adopted for a variety of visualizations tasks, but how far are we from perceptually aware LLMs that can predict human takeaways? Graphical perception literature has shown that human chart takeaways are sensitive to visualization design choices, such as spatial layouts. In this work, we examine the extent to which LLMs exhibit such sensitivity when generating takeaways, using bar charts with varying spatial layouts as a case study. We conducted three experiments and tested four common bar chart layouts: vertically juxtaposed, horizontally juxtaposed, overlaid, and stacked. In Experiment 1, we identified the optimal configurations to generate meaningful chart takeaways by testing four LLMs, two temperature settings, nine chart specifications, and two prompting strategies. We found that even state-of-the-art LLMs struggled to generate semantically diverse and factually accurate takeaways. In Experiment 2, we used the optimal configurations to generate 30 chart takeaways each for eight visualizations across four layouts and two datasets in both zero-shot and one-shot settings. Compared to human takeaways, we found that the takeaways LLMs generated often did not match the types of comparisons made by humans. In Experiment 3, we examined the effect of chart context and data on LLM takeaways. We found that LLMs, unlike humans, exhibited variation in takeaway comparison types for different bar charts using the same bar layout. Overall, our case study evaluates the ability of LLMs to emulate human interpretations of data and points to challenges and opportunities in using LLMs to predict human chart takeaways.

Fact-checking data claims requires data evidence retrieval and analysis, which can become tedious and intractable when done manually. This work presents Aletheia, an automated fact-checking prototype designed to facilitate data claims verification and enhance data evidence communication. For verification, we utilize a pre-trained LLM to parse the semantics for evidence retrieval. To effectively communicate the data evidence, we design representations in two forms: data tables and visualizations, tailored to various data fact types. Additionally, we design interactions that showcase a real-world application of these techniques. We evaluate the performance of two core NLP tasks with a curated dataset comprising 400 data claims and compare the two representation forms regarding viewers’ assessment time, confidence, and preference via a user study with 20 participants. The evaluation offers insights into the feasibility and bottlenecks of using LLMs for data fact-checking tasks, potential advantages and disadvantages of using visualizations over data tables, and design recommendations for presenting data evidence.

Machine Learning models for chart-grounded Q&A (CQA) often treat charts as images, but performing CQA on pixel values has proven challenging. We thus investigate a resource overlooked by current ML-based approaches: the declarative documents describing how charts should visually encode data (i.e., chart specifications). In this work, we use chart specifications to enhance language models (LMs) for chart-reading tasks, such that the resulting system can robustly understand language for CQA. Through a case study with 359 bar charts, we test novel fine tuning schemes on both GPT-3 and T5 using a new dataset curated for two CQA tasks: question-answering and visual explanation generation. Our text-only approaches strongly outperform vision-based GPT-4 on explanation generation (99% vs. 63% accuracy), and show promising results for question-answering (57–67% accuracy). Through in-depth experiments, we also show that our text-only approaches are mostly robust to natural language variation.

Designing responsive visualizations for various screen types can be tedious as authors must manage multiple chart versions across design iterations. Automated approaches for responsive visualization must take into account the user's need for agency in exploring possible design ideas and applying customizations based on their own goals. We design and implement Dupo, a mixedinitiative approach to creating responsive visualizations that combines the agency afforded by a manual interface with automation provided by a recommender system. Given an initial design, users can browse automated design suggestions for a different screen type and make edits to a chosen design, thereby supporting quick prototyping and customizability. Dupo employs a two-step recommender pipeline that first suggests significant design changes (Exploration) followed by more subtle changes (Alteration). We evaluated Dupo with six expert responsive visualization authors. While creating responsive versions of a source design in Dupo, participants could reason about different design suggestions without having to manually prototype them, and thus avoid prematurely fixating on a particular design. This process led participants to create designs that they were satisfied with but which they had previously overlooked.

Visual data stories can effectively convey insights from data, yet their creation often necessitates intricate data exploration, insight discovery, narrative organization, and customization to meet the communication objectives of the storyteller. Existing automated data storytelling techniques, however, tend to overlook the importance of user customization during the data story authoring process, limiting the system's ability to create tailored narratives that reflect the user's intentions. We present a novel data story generation workflow that leverages adaptive machine-guided elicitation of user feedback to customize the story. Our approach employs an adaptive plug-in module for existing story generation systems, which incorporates user feedback through interactive questioning based on the conversation history and dataset. This adaptability refines the system's understanding of the user's intentions, ensuring the final narrative aligns with their goals. We demonstrate the feasibility of our approach through the implementation of an interactive prototype: Socrates. Through a quantitative user study with 18 participants that compares our method to a state-of-the-art data story generation algorithm, we show that Socrates produces more relevant stories with a larger overlap of insights compared to human-generated stories. We also demonstrate the usability of Socrates via interviews with three data analysts and highlight areas of future work.

Visualization recommendation systems make understanding data more accessible to users of all skill levels by automatically generating visualizations for users to explore. However, most existing visualization recommendation systems focus on ranking all possible visualizations based on the attributes or encodings, which makes it difficult to find the most interesting or relevant insights. We therefore introduce a novel class of visualization recommendation systems that automatically rank and recommend both groups of related insights and the most important insights within each group. Our approach combines results across different learning-based methods to discover insights automatically and generalizes to a variety of attribute types (e.g., categorical, numerical, and temporal), including non-trivial combinations of these attribute types. We then implemented a new insight-centric visualization recommendation system, SpotLight, which ranks annotated visualizations in visual insight groups. Finally, we conducted a user study which showed that users are able to quickly understand and find relevant insights in unfamiliar data.

Modern organizations amass their datasets into centralized repositories called data lakes, affording analytics as needed. The resultant scale and complexity of these data lakes, however, can make data navigation and monitoring challenging for users. We present DataCockpit, a Python toolkit that leverages datasets, usage logs, and associated meta-data to provision data usage and quality characteristics. DataCockpit computes these characteristics for each attribute (e.g., number of times it was queried for subsequent use in downstream applications) and record (e.g., number of non-missing, valid values) and aggregates them at the level of datasets. We develop a visual monitoring tool, powered by DataCockpit, and demonstrate how it can assist data / system administrators as well as end-users to effectively navigate and monitor a data lake. DataCockpit and the monitoring tool are available as open source software for developers to build custom monitoring applications on top of data lakes.

Analyzing data and creating effective visualizations often requires extensive domain expertise. For users with less experience, it can be difficult to know how to get started with exploratory data analysis (EDA) and how to approach the code. Chatbots can reduce the gap between analysis outcomes and user expectations by leveraging multi-turn conversations to provide a more natural interface between the user and computer-agent. To inform the design of future visual EDA chatbots, we conduct a survey and interview study with ten potential users. Our results suggest that users want a visual EDA chatbot that can make exploratory data analysis easier, while also augmenting their knowledge of visualization and analysis techniques. Between the initial survey and post-interview questionnaire, we saw increased optimism overall for the usefulness and anticipated analytic ease of visual EDA chatbots. Based on these results, we identify four key design guidelines: future visual EDA chatbots should (1) understand the user’s data and intent, (2) respond with useful visualizations, (3) leverage the history of the visualizations and data, and (4) produce verifiable and shareable analysis processes.

Selecting relevant data subsets from large, unfamiliar datasets can be difficult. We address this challenge by modeling and visualizing two kinds of auxiliary information: (1) quality – the validity and appropriateness of data required to perform certain analytical tasks; and (2) usage – the historical utilization characteristics of data across multiple users. Through a design study with 14 data workers, we integrate this information into a visual data preparation and analysis tool, DataPilot. DataPilot presents visual cues about “the good, the bad, and the ugly” aspects of data and provides graphical user interface controls as interaction affordances, guiding users to perform subset selection. Through a study with 36 participants, we investigate how DataPilot helps users navigate a large, unfamiliar tabular dataset, prepare a relevant subset, and build a visualization dashboard. We find that users selected smaller, effective subsets with higher quality and usage, and with greater success and confidence.

While paper instructions are a mainstream medium for sharing knowledge, consuming such instructions and translating them into activities can be inefficient due to the lack of connectivity with the physical environment. We propose PaperToPlace, a novel workflow comprising an authoring pipeline, which allows the authors to rapidly transform and spatialize existing paper instructions into an MR experience, and a consumption pipeline, which computationally places each instruction step at an optimal location that is easy to read and does not occlude key interaction areas. Our evaluation of the authoring pipeline with 12 participants demonstrates the usability of our workflow and the effectiveness of using a machine learning based approach to help extract the spatial locations associated with each step. A second within-subjects study with another 12 participants demonstrates the merits of our consumption pipeline to reduce context-switching effort by delivering individual segmented instruction steps and offering hands-free affordances.

Computational notebooks such as Jupyter are popular for exploratory data analysis and insight finding. Despite the module-based structure, notebooks visually appear as a single thread of interleaved cells containing text, code, visualizations, and tables, which can be unorganized and obscure users' data analysis workflow. Furthermore, users with limited coding expertise may struggle to quickly engage in the analysis process. In this work, we design and implement an interactive notebook framework, WhatsNext, with the goal of supporting low-code visual data exploration with insight-based user guidance. In particular, we (1) re-design a standard notebook cell to include a recommendation panel that suggests possible next-step exploration questions or analysis actions to take, and (2) create an interactive, dynamic tree visualization that reflects the analytic dependencies between notebook cells to make it easy for users to see the structure of the data exploration threads and trace back to previous steps.

The language for expressing comparisons is often complex and nuanced, making supporting natural language-based visual comparison a non-trivial task. To better understand how people reason about comparisons in natural language, we explore a design space of utterances for comparing data entities. We identified different parameters of comparison utterances that indicate what is being compared (i.e., data variables and attributes) as well as how these parameters are specified (i.e., explicitly or implicitly). We conducted a user study with sixteen data visualization experts and non-experts to investigate how they designed visualizations for comparisons in our design space. Based on the rich set of visualization techniques observed, we extracted key design features from the visualizations and synthesized them into a subset of sixteen representative visualization designs. We then conducted a follow-up study to validate user preferences for the sixteen representative visualizations corresponding to utterances in our design space. Findings from these studies suggest guidelines and future directions for designing natural language interfaces and recommendation tools to better support natural language comparisons in visual analytics.

Most real-world datasets contain missing values yet most exploratory data analysis (EDA) systems only support visualising data points with complete cases. This omission may potentially lead the user to biased analyses and insights. Imputation techniques can help estimate the value of a missing data point, but introduces additional uncertainty. In this work, we investigate the effects of visualising imputed values in charts using different ways of representing data imputations and imputation uncertainty-no imputation, mean, 95% confidence intervals, probability density plots, gradient intervals, and hypothetical outcome plots. We focus on scatterplots, which is a commonly used chart type, and conduct a crowdsourced study with 202 participants. We measure users' bias and precision in performing two tasks-estimating average and detecting trend-and their self-reported confidence in performing these tasks. Our results suggest that, when estimating averages, uncertainty representations may reduce bias but at the cost of decreasing precision. When estimating trend, only hypothetical outcome plots may lead to a small probability of reducing bias while increasing precision. Participants in every uncertainty representation were less certain about their response when compared to the baseline. The findings point towards potential trade-offs in using uncertainty encodings for datasets with a large number of missing values. This paper and the associated analysis materials are available at: https://osf.io/q4y5r/.

Designing responsive visualizations can be cast as applying transformations to a source view to render it suitable for a different screen size. However, designing responsive visualizations is often tedious as authors must manually apply and reason about candidate transformations. We present Cicero, a declarative grammar for concisely specifying responsive visualization transformations which paves the way for more intelligent responsive visualization authoring tools. Cicero’s flexible specifier syntax allows authors to select visualization elements to transform, independent of the source view’s structure. Cicero encodes a concise set of actions to encode a diverse set of transformations in both desktop-first and mobile-first design processes. Authors can ultimately reuse design-agnostic transformations across different visualizations. To demonstrate the utility of Cicero, we develop a compiler to an extended version of Vega-Lite, and provide principles for our compiler. We further discuss the incorporation of Cicero into responsive visualization authoring tools, such as a design recommender.

Efficient outage detection and remediation is crucial for effectively operating cloud computing systems. To remediate outages, system engineers must quickly identify the causal relationships between metrics and correlate events across multiple monitoring tools. In practice, this process largely remains reactive due to the complexity and general lack of interpretability within such monitoring environments. This work presents ViSRE: an integrated visual analytics system that integrates causal and predictive models with interactive visualizations to aid in proactive cloud outage management. We develop enhanced node representations for our causal graph representation to support system engineers in performing root cause analysis and reasoning about causality chains in multi-dimensional temporal data. We report the results of a quantitative assessment of the proposed predictive models, which show good performance guarantees. To evaluate and refine our system, we conduct a study with six cloud system engineers who verify that our proposed techniques can support proactive cloud maintenance by intuitively displaying temporal relationships between predicted and raw data. By correlating and presenting data from disparate sources, ViSRE also reduces context switching costs and reduces the time spent on manually correlating events during remediation of time-critical outages.

In this work, we develop a Graph Neural Network (GNN) framework for the problem of personalized visualization recommendation. The GNN-based framework first represents the large corpus of datasets and visualizations from users as a large heterogeneous graph. Then, it decomposes a visualization into its data and visual components, and then jointly models each of them as a large graph to obtain embeddings of the users, attributes (across all datasets in the corpus), and visual-configurations. From these user-specific embeddings of the attributes and visual-configurations, we can predict the probability of any visualization arising from a specific user. Finally, the experiments demonstrated the effectiveness of using graph neural networks for automatic and personalized recommendation of visualizations to specific users based on their data and visual (design choice) preferences. To the best of our knowledge, this is the first such work to develop and leverage GNNs for this problem.

Although we have seen a proliferation of algorithms for recommending visualizations, these algorithms are rarely compared with one another, making it difficult to ascertain which algorithm is best for a given visual analysis scenario. Though several formal frameworks have been proposed in response, we believe this issue persists because visualization recommendation algorithms are inadequately specified from an evaluation perspective. In this paper, we propose an evaluation-focused framework to contextualize and compare a broad range of visualization recommendation algorithms. We present the structure of our framework, where algorithms are specified using three components: (1) a graph representing the full space of possible visualization designs, (2) the method used to traverse the graph for potential candidates for recommendation, and (3) an oracle used to rank candidate designs. To demonstrate how our framework guides the formal comparison of algorithmic performance, we not only theoretically compare five existing representative recommendation algorithms, but also empirically compare four new algorithms generated based on our findings from the theoretical comparison. Our results show that these algorithms behave similarly in terms of user performance, highlighting the need for more rigorous formal comparisons of recommendation algorithms to further clarify their benefits in various analysis scenarios.

Responsive visualizations adapt to effectively present information based on the device context. Such adaptations are essential for news content that is increasingly consumed on mobile devices. However, existing tools provide little support for responsive visualization design. We analyze a corpus of 231 responsive news visualizations and discuss formative interviews with five journalists about responsive visualization design. These interviews motivate four central design guidelines: enable simultaneous cross-device edits, facilitate device-specific customization, show cross-device previews, and support propagation of edits. Based on these guidelines, we present a prototype system that allows users to preview and edit multiple visualization versions simultaneously. We demonstrate the utility of the system features by recreating four real-world responsive visualizations from our corpus.

PDF documents often contain rich data tables that offer opportunities for dynamic reuse in new interactive applications. We describe a pipeline for extracting, analyzing, and parsing PDF tables based on existing machine learning and rule-based techniques. Implementing and deploying this pipeline on a corpus of 447 documents with 1,171 tables results in only 11 tables that are correctly extracted and parsed. To improve the results of automatic table analysis, we first present a taxonomy of errors that arise in the analysis pipeline and discuss the implications of cascading errors on the user experience. We then contribute a system with two sets of lightweight interaction techniques (gesture and toolbar), for viewing and repairing extraction errors in PDF tables on mobile devices. In an evaluation with 17 users involving both a phone and a tablet, participants effectively repaired common errors in 10 tables, with an average time of about 2 minutes per table.

Programmers must draw explicit connections between their code and runtime state to properly assess the correctness of their programs. However, debugging tools often decouple the program state from the source code and require explicitly invoked views to bridge the rift between program editing and program understanding. To unobtrusively reveal runtime behavior during both normal execution and debugging, we contribute techniques for visualizing program variables directly within the source code. We describe a design space and placement criteria for embedded visualizations. We evaluate our in situ visualizations in an editor for the Vega visualization grammar. Compared to a baseline development environment, novice Vega users improve their overall task grade by about 2 points when using the in situ visualizations and exhibit significant positive effects on their self-reported speed and accuracy.

Abstract Constraints enable flexible graph layout by combining the ease of automatic layout with customizations for a particular domain. However, constraint‐based layout often requires many individual constraints defined over specific nodes and node pairs. In addition to the effort of writing and maintaining a large number of similar constraints, such constraints are specific to the particular graph and thus cannot generalize to other graphs in the same domain. To facilitate the specification of customized and generalizable constraint layouts, we contribute SetCoLa: a domain‐specific language for specifying high‐level constraints relative to properties of the backing data. Users identify node sets based on data or graph properties and apply high‐level constraints within each set. Applying constraints to node sets rather than individual nodes reduces specification effort and facilitates reapplication of customized layouts across distinct graphs. We demonstrate the conciseness, generalizability, and expressiveness of SetCoLa on a series of real‐world examples from ecological networks, biological systems, and social networks.

Patient-generated data can allow patients and providers to collaboratively develop accurate diagnoses and actionable treatment plans. Unfortunately, patients and providers often lack effective support to make use of such data. We examine patient-provider collaboration to interpret patient-generated data. We focus on irritable bowel syndrome (IBS), a chronic illness in which particular foods can exacerbate symptoms. IBS management often requires patient-provider collaboration using a patient's food and symptom journal to identify the patient's triggers. We contribute interactive visualizations to support exploration of such journals, as well as an examination of patient-provider collaboration in interpreting the journals. Drawing upon individual and collaborative interviews with patients and providers, we find that collaborative review helps improve data comprehension and build mutual trust. We also find a desire to use tools like our interactive visualizations within and beyond clinic appointments. We discuss these findings and present guidance for the design of future tools.

Abstract Interaction is critical to effective visualization, but can be difficult to author and debug due to dependencies among input events, program state, and visual output. Recent advances leverage reactive semantics to support declarative design and avoid the “spaghetti code” of imperative event handlers. While reactive programming improves many aspects of development, textual specifications still fail to convey the complex runtime dynamics. In response, we contribute a set of visual debugging techniques to reveal the runtime behavior of reactive visualizations. A timeline view records input events and dynamic variable updates, allowing designers to replay and inspect the propagation of values step‐by‐step. On‐demand annotations overlay the output visualization to expose relevant state and scale mappings in‐situ. Dynamic tables visualize how backing datasets change over time. To evaluate the effectiveness of these techniques, we study how first‐time Vega users debug interactions in faulty, unfamiliar specifications; with no prior knowledge, participants were able to accurately trace errors through the specification.

We present Reactive Vega, a system architecture that provides the first robust and comprehensive treatment of declarative visual and interaction design for data visualization. Starting from a single declarative specification, Reactive Vega constructs a dataflow graph in which input data, scene graph elements, and interaction events are all treated as first-class streaming data sources. To support expressive interactive visualizations that may involve time-varying scalar, relational, or hierarchical data, Reactive Vega's dataflow graph can dynamically re-write itself at runtime by extending or pruning branches in a data-driven fashion. We discuss both compile- and run-time optimizations applied within Reactive Vega, and share the results of benchmark studies that indicate superior interactive performance to both D3 and the original, non-reactive Vega system.