Search

Creative writers have a love for their craft, yet AI systems using large language models (LLMs) offer the automation of significant parts of the writing process. So why do some creative writers choose to integrate AI into their workflows? To explore this, we interview and observe a writing session with 18 creative writers who already use AI regularly in their writing practice. Our findings reveal that creative writers are intentional about how they incorporate AI, making many deliberate decisions about when and how to engage AI based on the core values they hold about writing. These values, such as authenticity and craftsmanship, alongside writers' relationships with and use of AI influence the parts of writing over which they wish to maintain control. Through our analysis, we contribute a taxonomy of writer values, writer relationships with AI, and integration strategies, and discuss how these three elements interrelate.

Data-driven scientific discovery requires the iterative integration of scientific domain knowledge, statistical expertise, and an understanding of data semantics to make nuanced analytical decisions, e.g., about which variables, transformations, and statistical models to consider. LM-based agents equipped with planning, memory, and code execution capabilities have the potential to support data-driven science. However, evaluating agents on such open-ended tasks is challenging due to multiple valid approaches, partially correct steps, and different ways to express the same decisions. To address these challenges, we present BLADE, a benchmark to automatically evaluate agents' multifaceted approaches to open-ended research questions. BLADE consists of 12 datasets and research questions drawn from existing scientific literature, with ground truth collected from independent analyses by expert data scientists and researchers. To automatically evaluate agent responses, we developed corresponding computational methods to match different representations of analyses to this ground truth. Though language models possess considerable world knowledge, our evaluation shows that they are often limited to basic analyses. However, agents capable of interacting with the underlying data demonstrate improved, but still non-optimal, diversity in their analytical decision making. Our work enables the evaluation of agents for data-driven science and provides researchers deeper insights into agents' analysis approaches.

Trained on vast corpora, Large Language Models (LLMs) have the potential to encode visualization design knowledge and best practices. However, if they fail to do so, they might provide unreliable visualization recommendations. What visualization design preferences, then, have LLMs learned? We contribute DracoGPT, a method for extracting, modeling, and assessing visualization design preferences from LLMs. To assess varied tasks, we develop two pipelines--DracoGPT-Rank and DracoGPT-Recommend--to model LLMs prompted to either rank or recommend visual encoding specifications. We use Draco as a shared knowledge base in which to represent LLM design preferences and compare them to best practices from empirical research. We demonstrate that DracoGPT can accurately model the preferences expressed by LLMs, enabling analysis in terms of Draco design constraints. Across a suite of backing LLMs, we find that DracoGPT-Rank and DracoGPT-Recommend moderately agree with each other, but both substantially diverge from guidelines drawn from human subjects experiments. Future work can build on our approach to expand Draco's knowledge base to model a richer set of preferences and to provide a robust and cost-effective stand-in for LLMs., Comment: IEEE Transactions on Visualization and Computer Graphics (Proc. VIS 2024)

Visualization linters are end-user facing evaluators that automatically identify potential chart issues. These spell-checker like systems offer a blend of interpretability and customization that is not found in other forms of automated assistance. However, existing linters do not model context and have primarily targeted users who do not need assistance, resulting in obvious -- even annoying -- advice. We investigate these issues within the domain of color palette design, which serves as a microcosm of visualization design concerns. We contribute a GUI-based color palette linter as a design probe that covers perception, accessibility, context, and other design criteria, and use it to explore visual explanations, integrated fixes, and user defined linting rules. Through a formative interview study and theory-driven analysis, we find that linters can be meaningfully integrated into graphical contexts thereby addressing many of their core issues. We discuss implications for integrating linters into visualization tools, developing improved assertion languages, and supporting end-user tunable advice -- all laying the groundwork for more effective visualization linters in any context., Comment: To appear at VIS2024

Geovisualizations are powerful tools for exploratory spatial analysis, enabling sighted users to discern patterns, trends, and relationships within geographic data. However, these visual tools have remained largely inaccessible to screen-reader users. We present AltGeoViz, a new system we designed to facilitate geovisualization exploration for these users. AltGeoViz dynamically generates alt-text descriptions based on the user's current map view, providing summaries of spatial patterns and descriptive statistics. In a study of five screen-reader users, we found that AltGeoViz enabled them to interact with geovisualizations in previously infeasible ways. Participants demonstrated a clear understanding of data summaries and their location context, and they could synthesize spatial understandings of their explorations. Moreover, we identified key areas for improvement, such as the addition of intuitive spatial navigation controls and comparative analysis features.

Data analysts have long sought to turn unstructured text data into meaningful concepts. Though common, topic modeling and clustering focus on lower-level keywords and require significant interpretative work. We introduce concept induction, a computational process that instead produces high-level concepts, defined by explicit inclusion criteria, from unstructured text. For a dataset of toxic online comments, where a state-of-the-art BERTopic model outputs "women, power, female," concept induction produces high-level concepts such as "Criticism of traditional gender roles" and "Dismissal of women's concerns." We present LLooM, a concept induction algorithm that leverages large language models to iteratively synthesize sampled text and propose human-interpretable concepts of increasing generality. We then instantiate LLooM in a mixed-initiative text analysis tool, enabling analysts to shift their attention from interpreting topics to engaging in theory-driven analysis. Through technical evaluations and four analysis scenarios ranging from literature review to content moderation, we find that LLooM's concepts improve upon the prior art of topic models in terms of quality and data coverage. In expert case studies, LLooM helped researchers to uncover new insights even from familiar datasets, for example by suggesting a previously unnoticed concept of attacks on out-party stances in a political social media dataset., Comment: To appear at CHI 2024

The ubiquity and on-the-go availability of mobile devices makes them central to many tasks such as interpersonal communication and media consumption. However, despite the potential of mobile devices for on-demand exploratory data visualization, existing mobile interactions are difficult, often using highly custom interactions, complex gestures, or multi-modal input. We synthesize limitations from the literature and outline four motivating principles for improved mobile interaction: leverage ubiquitous modalities, prioritize discoverability, enable rapid in-context data exploration, and promote graceful recovery. We then contribute thirteen interaction candidates and conduct a formative study with twelve participants who experienced our interactions in a testbed prototype. Based on these interviews, we discuss design considerations and tradeoffs from four main themes: precise and rapid inspection, focused navigation, single-touch and fixed orientation interaction, and judicious use of motion., Comment: 4 pages, 1 figure, 1 table, EuroVis 2024 Short Papers

LLM chains enable complex tasks by decomposing work into a sequence of subtasks. Similarly, the more established techniques of crowdsourcing workflows decompose complex tasks into smaller tasks for human crowdworkers. Chains address LLM errors analogously to the way crowdsourcing workflows address human error. To characterize opportunities for LLM chaining, we survey 107 papers across the crowdsourcing and chaining literature to construct a design space for chain development. The design space covers a designer's objectives and the tactics used to build workflows. We then surface strategies that mediate how workflows use tactics to achieve objectives. To explore how techniques from crowdsourcing may apply to chaining, we adapt crowdsourcing workflows to implement LLM chains across three case studies: creating a taxonomy, shortening text, and writing a short story. From the design space and our case studies, we identify takeaways for effective chain design and raise implications for future research and development.

Dynamically Interactive Visualization (DIVI) is a novel approach for orchestrating interactions within and across static visualizations. DIVI deconstructs Scalable Vector Graphics charts at runtime to infer content and coordinate user input, decoupling interaction from specification logic. This decoupling allows interactions to extend and compose freely across different tools, chart types, and analysis goals. DIVI exploits positional relations of marks to detect chart components such as axes and legends, reconstruct scales and view encodings, and infer data fields. DIVI then enumerates candidate transformations across inferred data to perform linking between views. To support dynamic interaction without prior specification, we introduce a taxonomy that formalizes the space of standard interactions by chart element, interaction type, and input event. We demonstrate DIVI's usefulness for rapid data exploration and analysis through a usability study with 13 participants and a diverse gallery of dynamically interactive visualizations, including single chart, multi-view, and cross-tool configurations., Comment: 9 pages, 2 pages supplementary material, 10 figures, IEEE TVCG 2024 (Proc. VIS 2023)

Statistical models should accurately reflect analysts' domain knowledge about variables and their relationships. While recent tools let analysts express these assumptions and use them to produce a resulting statistical model, it remains unclear what analysts want to express and how externalization impacts statistical model quality. This paper addresses these gaps. We first conduct an exploratory study of analysts using a domain-specific language (DSL) to express conceptual models. We observe a preference for detailing how variables relate and a desire to allow, and then later resolve, ambiguity in their conceptual models. We leverage these findings to develop rTisane, a DSL for expressing conceptual models augmented with an interactive disambiguation process. In a controlled evaluation, we find that rTisane's DSL helps analysts engage more deeply with and accurately externalize their assumptions. rTisane also leads to statistical models that match analysts' assumptions, maintain analysis intent, and better fit the data.

Data analysis is challenging as analysts must navigate nuanced decisions that may yield divergent conclusions. AI assistants have the potential to support analysts in planning their analyses, enabling more robust decision making. Though AI-based assistants that target code execution (e.g., Github Copilot) have received significant attention, limited research addresses assistance for both analysis execution and planning. In this work, we characterize helpful planning suggestions and their impacts on analysts' workflows. We first review the analysis planning literature and crowd-sourced analysis studies to categorize suggestion content. We then conduct a Wizard-of-Oz study (n=13) to observe analysts' preferences and reactions to planning assistance in a realistic scenario. Our findings highlight subtleties in contextual factors that impact suggestion helpfulness, emphasizing design implications for supporting different abstractions of assistance, forms of initiative, increased engagement, and alignment of goals between analysts and assistants., Comment: Accepted to CHI 2024

Findings from graphical perception can guide visualization recommendation algorithms in identifying effective visualization designs. However, existing algorithms use knowledge from, at best, a few studies, limiting our understanding of how complementary (or contradictory) graphical perception results influence generated recommendations. In this paper, we present a pipeline of applying a large body of graphical perception results to develop new visualization recommendation algorithms and conduct an exploratory study to investigate how results from graphical perception can alter the behavior of downstream algorithms. Specifically, we model graphical perception results from 30 papers in Draco -- a framework to model visualization knowledge -- to develop new recommendation algorithms. By analyzing Draco-generated algorithms, we showcase the feasibility of our method to (1) identify gaps in existing graphical perception literature informing recommendation algorithms, (2) cluster papers by their preferred design rules and constraints, and (3) investigate why certain studies can dominate Draco's recommendations, whereas others may have little influence. Given our findings, we discuss the potential for mutually reinforcing advancements in graphical perception and visualization recommendation research.

Visual analytics (VA) tools support data exploration by helping analysts quickly and iteratively generate views of data which reveal interesting patterns. However, these tools seldom enable explicit checks of the resulting interpretations of data -- e.g., whether patterns can be accounted for by a model that implies a particular structure in the relationships between variables. We present EVM, a data exploration tool that enables users to express and check provisional interpretations of data in the form of statistical models. EVM integrates support for visualization-based model checks by rendering distributions of model predictions alongside user-generated views of data. In a user study with data scientists practicing in the private and public sector, we evaluate how model checks influence analysts' thinking during data exploration. Our analysis characterizes how participants use model checks to scrutinize expectations about data generating process and surfaces further opportunities to scaffold model exploration in VA tools.

A multiverse analysis evaluates all combinations of "reasonable" analytic decisions to promote robustness and transparency, but can lead to a combinatorial explosion of analyses to compute. Long delays before assessing results prevent users from diagnosing errors and iterating early. We contribute (1) approximation algorithms for estimating multiverse sensitivity and (2) monitoring visualizations for assessing progress and controlling execution on the fly. We evaluate how quickly three sampling-based algorithms converge to accurately rank sensitive decisions in both synthetic and real multiverse analyses. Compared to uniform random sampling, round robin and sketching approaches are 2 times faster in the best case, while on average estimating sensitivity accurately using 20% of the full multiverse. To enable analysts to stop early to fix errors or decide when results are "good enough" to move forward, we visualize both effect size and decision sensitivity estimates with confidence intervals, and surface potential issues including runtime warnings and model quality metrics.

Efforts to push programming beyond static textual code have sought to imbue programming with multiple distinct qualities. One long-acknowledged quality is liveness: providing programmers with in-depth feedback about a program's dynamic behavior as the program is edited. A second quality, long-explored but lacking a shared term of art, is richness: allowing programmers to edit programs though domain-specific representations and interactions rather than solely through text. In this paper, we map the relationship between these two qualities and survey past work that exemplifies them. We observe that systems combining liveness and richness often do so at the cost of an essential quality of traditional programming: composability. We argue that, by combining liveness, richness, and composability, programming systems can better capture the full potential of interactive computation without leaving behind the expressivity of traditional code., Comment: To appear in the proceedings of PLATEAU 2023

The in-context learning capabilities of LLMs like GPT-3 allow annotators to customize an LLM to their specific tasks with a small number of examples. However, users tend to include only the most obvious patterns when crafting examples, resulting in underspecified in-context functions that fall short on unseen cases. Further, it is hard to know when "enough" examples have been included even for known patterns. In this work, we present ScatterShot, an interactive system for building high-quality demonstration sets for in-context learning. ScatterShot iteratively slices unlabeled data into task-specific patterns, samples informative inputs from underexplored or not-yet-saturated slices in an active learning manner, and helps users label more efficiently with the help of an LLM and the current example set. In simulation studies on two text perturbation scenarios, ScatterShot sampling improves the resulting few-shot functions by 4-5 percentage points over random sampling, with less variance as more examples are added. In a user study, ScatterShot greatly helps users in covering different patterns in the input space and labeling in-context examples more efficiently, resulting in better in-context learning and less user effort., Comment: IUI 2023: 28th International Conference on Intelligent User Interfaces

The many genres of narrative visualization (e.g. data comics, data videos) each offer a unique set of affordances and constraints. To better understand a genre that we call cinematic visualizations-3D visualizations that make highly deliberate use of a camera to convey a narrative-we gathered 50 examples and analyzed their traditional cinematic aspects to identify the benefits and limitations of the form. While the cinematic visualization approach can violate traditional rules of visualization, we find that through careful control of the camera, cinematic visualizations enable immersion in data-driven, anthropocentric environments, and can naturally incorporate in-situ narrators, concrete scales, and visual analogies. Our analysis guides our design of a series of cinematic visualizations, created for NASA's Earth Science Communications team. We present one as a case study to convey design guidelines covering cinematography, lighting, set design, and sound, and discuss challenges in creating cinematic visualizations., Comment: 10 pages, 7 figures

Narrative visualization is a powerful communicative tool that can take on various formats such as interactive articles, slideshows, and data videos. These formats each have their strengths and weaknesses, but existing authoring tools only support one output target. We conducted a series of formative interviews with seven domain experts to understand needs and practices around cross-format data stories, and developed Fidyll, a cross-format compiler for authoring interactive data stories and explorable explanations. Our open-source tool can be used to rapidly create formats including static articles, low-motion articles, interactive articles, slideshows, and videos. We evaluate our system through a series of real-world usage scenarios, showing how it benefits authors in the domains of data journalism, scientific publishing, and nonprofit advocacy. We show how Fidyll, provides expressive leverage by reducing the amount of non-narrative markup that authors need to write by 80-90% compared to Idyll, an existing markup language for authoring interactive articles., Comment: 10 pages, 6 figures, for associated examples see https://idyll-lang.github.io/fidyll-examples/

Proper statistical modeling incorporates domain theory about how concepts relate and details of how data were measured. However, data analysts currently lack tool support for recording and reasoning about domain assumptions, data collection, and modeling choices in an integrated manner, leading to mistakes that can compromise scientific validity. For instance, generalized linear mixed-effects models (GLMMs) help answer complex research questions, but omitting random effects impairs the generalizability of results. To address this need, we present Tisane, a mixed-initiative system for authoring generalized linear models with and without mixed-effects. Tisane introduces a study design specification language for expressing and asking questions about relationships between variables. Tisane contributes an interactive compilation process that represents relationships in a graph, infers candidate statistical models, and asks follow-up questions to disambiguate user queries to construct a valid model. In case studies with three researchers, we find that Tisane helps them focus on their goals and assumptions while avoiding past mistakes.

Complex animated transitions may be easier to understand when divided into separate, consecutive stages. However, effective staging requires careful attention to both animation semantics and timing parameters. We present Gemini2, a system for creating staged animations from a sequence of chart keyframes. Given only a start state and an end state, Gemini2 can automatically recommend intermediate keyframes for designers to consider. The Gemini2 recommendation engine leverages Gemini, our prior work, and GraphScape to itemize the given complex change into semantic edit operations and to recombine operations into stages with a guided order for clearly conveying the semantics. To evaluate Gemini2's recommendations, we conducted a human-subject study in which participants ranked recommended animations from both Gemini2 and Gemini. We find that Gemini2's animation recommendation ranking is well aligned with subjects' preferences, and Gemini2 can recommend favorable animations that Gemini cannot support.

Data analysis requires translating higher level questions and hypotheses into computable statistical models. We present a mixed-methods study aimed at identifying the steps, considerations, and challenges involved in operationalizing hypotheses into statistical models, a process we refer to as hypothesis formalization. In a formative content analysis of research papers, we find that researchers highlight decomposing a hypothesis into sub-hypotheses, selecting proxy variables, and formulating statistical models based on data collection design as key steps. In a lab study, we find that analysts fixated on implementation and shaped their analysis to fit familiar approaches, even if sub-optimal. In an analysis of software tools, we find that tools provide inconsistent, low-level abstractions that may limit the statistical models analysts use to formalize hypotheses. Based on these observations, we characterize hypothesis formalization as a dual-search process balancing conceptual and statistical considerations constrained by data and computation, and discuss implications for future tools.

While counterfactual examples are useful for analysis and training of NLP models, current generation methods either rely on manual labor to create very few counterfactuals, or only instantiate limited types of perturbations such as paraphrases or word substitutions. We present Polyjuice, a general-purpose counterfactual generator that allows for control over perturbation types and locations, trained by finetuning GPT-2 on multiple datasets of paired sentences. We show that Polyjuice produces diverse sets of realistic counterfactuals, which in turn are useful in various distinct applications: improving training and evaluation on three different tasks (with around 70% less annotation effort than manual generation), augmenting state-of-the-art explanation techniques, and supporting systematic counterfactual error analysis by revealing behaviors easily missed by human experts., Comment: ACL 2021, main conference, long paper

Animated transitions help viewers follow changes between related visualizations. Specifying effective animations demands significant effort: authors must select the elements and properties to animate, provide transition parameters, and coordinate the timing of stages. To facilitate this process, we present Gemini, a declarative grammar and recommendation system for animated transitions between single-view statistical graphics. Gemini specifications define transition "steps" in terms of high-level visual components (marks, axes, legends) and composition rules to synchronize and concatenate steps. With this grammar, Gemini can recommend animation designs to augment and accelerate designers' work. Gemini enumerates staged animation designs for given start and end states, and ranks those designs using a cost function informed by prior perceptual studies. To evaluate Gemini, we conduct both a formative study on Mechanical Turk to assess and tune our ranking function, and a summative study in which 8 experienced visualization developers implement animations in D3 that we then compare to Gemini's suggestions. We find that most designs (9/11) are exactly replicable in Gemini, with many (8/11) achievable via edits to suggestions, and that Gemini suggestions avoid multiple participant errors.

Large scale analysis of source code, and in particular scientific source code, holds the promise of better understanding the data science process, identifying analytical best practices, and providing insights to the builders of scientific toolkits. However, large corpora have remained unanalyzed in depth, as descriptive labels are absent and require expert domain knowledge to generate. We propose a novel weakly supervised transformer-based architecture for computing joint representations of code from both abstract syntax trees and surrounding natural language comments. We then evaluate the model on a new classification task for labeling computational notebook cells as stages in the data analysis process from data import to wrangling, exploration, modeling, and evaluation. We show that our model, leveraging only easily-available weak supervision, achieves a 38% increase in accuracy over expert-supplied heuristics and outperforms a suite of baselines. Our model enables us to examine a set of 118,000 Jupyter Notebooks to uncover common data analysis patterns. Focusing on notebooks with relationships to academic articles, we conduct the largest ever study of scientific code and find that notebook composition correlates with the citation count of corresponding papers.

Multiverse analysis is an approach to data analysis in which all "reasonable" analytic decisions are evaluated in parallel and interpreted collectively, in order to foster robustness and transparency. However, specifying a multiverse is demanding because analysts must manage myriad variants from a cross-product of analytic decisions, and the results require nuanced interpretation. We contribute Boba: an integrated domain-specific language (DSL) and visual analysis system for authoring and reviewing multiverse analyses. With the Boba DSL, analysts write the shared portion of analysis code only once, alongside local variations defining alternative decisions, from which the compiler generates a multiplex of scripts representing all possible analysis paths. The Boba Visualizer provides linked views of model results and the multiverse decision space to enable rapid, systematic assessment of consequential decisions and robustness, including sampling uncertainty and model fit. We demonstrate Boba's utility through two data analysis case studies, and reflect on challenges and design opportunities for multiverse analysis software., Comment: submitted to IEEE Transactions on Visualization and Computer Graphics (Proc. VAST)

How do analysis goals and context affect exploratory data analysis (EDA)? To investigate this question, we conducted semi-structured interviews with 18 data analysts. We characterize common exploration goals: profiling (assessing data quality) and discovery (gaining new insights). Though the EDA literature primarily emphasizes discovery, we observe that discovery only reliably occurs in the context of open-ended analyses, whereas all participants engage in profiling across all of their analyses. We describe the process and challenges of EDA highlighted by our interviews. We find that analysts must perform repetitive tasks (e.g., examine numerous variables), yet they may have limited time or lack domain knowledge to explore data. Analysts also often have to consult other stakeholders and oscillate between exploration and other tasks, such as acquiring and wrangling additional data. Based on these observations, we identify design opportunities for exploratory analysis tools, such as augmenting exploration with automation and guidance.

Drawing reliable inferences from data involves many, sometimes arbitrary, decisions across phases of data collection, wrangling, and modeling. As different choices can lead to diverging conclusions, understanding how researchers make analytic decisions is important for supporting robust and replicable analysis. In this study, we pore over nine published research studies and conduct semi-structured interviews with their authors. We observe that researchers often base their decisions on methodological or theoretical concerns, but subject to constraints arising from the data, expertise, or perceived interpretability. We confirm that researchers may experiment with choices in search of desirable results, but also identify other reasons why researchers explore alternatives yet omit findings. In concert with our interviews, we also contribute visualizations for communicating decision processes throughout an analysis. Based on our results, we identify design opportunities for strengthening end-to-end analysis, for instance via tracking and meta-analysis of multiple decision paths.

An emerging generation of visualization authoring systems support expressive information visualization without textual programming. As they vary in their visualization models, system architectures, and user interfaces, it is challenging to directly compare these systems using traditional evaluative methods. Recognizing the value of contextualizing our decisions in the broader design space, we present critical reflections on three systems we developed -- Lyra, Data Illustrator, and Charticulator. This paper surfaces knowledge that would have been daunting within the constituent papers of these three systems. We compare and contrast their (previously unmentioned) limitations and trade-offs between expressivity and learnability. We also reflect on common assumptions that we made during the development of our systems, thereby informing future research directions in visualization authoring systems.

In this paper, we describe a research agenda for deriving design principles directly from data. We argue that it is time to go beyond manually curated and applied visualization design guidelines. We propose learning models of visualization design from data collected using graphical perception studies and build tools powered by the learned models. To achieve this vision, we need to 1) develop scalable methods for collecting training data, 2) collect different forms of training data, 3) advance interpretability of machine learning models, and 4) develop adaptive models that evolve as more data becomes available.