Science AI Journal
Computer Science · Research topic

Open research questions in Data Visualization and Analytics

31 unresolved questions extracted from the limitations and future-work sections of 1,192 Data Visualization and Analytics papers in our library. Each links back to the study that raised it.

What the literature leaves open

  • These preliminary findings should be further explored in future research, particularly with systematic manipulation of data characteristics using a large number of SCED graphs.

    Exploring the potential of Generative AI as a support tool for single-case data analysis · 2026 · DOI

  • However, literature has shown mixed findings regarding agreement among visual analysts and inaccurate conclusions drawn from visual analysis (Dart & Radley, 2024; Fisher et al.

    Exploring the potential of Generative AI as a support tool for single-case data analysis · 2026 · DOI

  • Future research could explore whether more experienced practitioners or alter- native training methods yield different results, particularly in distinguishing between baseline trend and intervention effects. Limitations of this study include the focus on pre-service teachers enrolled in a specific educational program, which might not generalize to other populations or experienced practitioners.

    Enhancing Pre-service Teachers’ Visual Analysis Skills for Single-Case Graphs: the Role of Trend and Intervention Effect · 2026 · DOI

  • Our findings are bounded to the experimental regime we evaluated. First, we tested dense AMs of moderate size and a fixed cell resolu- tion; performance may differ for larger/sparser networks or smaller cells. Second, we relied on a single real-world dataset and simpli- fied the data to an undirected network; other domains, distributions, and directed networks may yield different trade-offs. Third, we did not separately optimize each encoding for optimal parameters; bet- ter per-encoding optimization could change performance. Fourth, we focus on two quantitative edge attributes summarized as central tendency and dispersion (i.e., mean/std); results may not transfer to other summary pairs (e.g., median/IQR), more than two attributes, or non-quantitative attributes. Finally, our task set is representative but not exhaustive; additional analytical tasks could reveal different results. Future work should extend this benchmark across matrix sizes/densities and cell resolutions, include directed networks and additional datasets/domains, test alternative distribution summaries (e.g., median/IQR) and more-than-bivariate attributes, systemati- cally tune each encoding, and broaden the task set.

    Evaluating Encodings for Bivariate Edges in Adjacency Matrices · 2026 · DOI

  • Several limitations should be acknowledged. An important limitation concerns the instruc- tional design of the intervention itself. The training examined in this study consisted of an intensive initial instructional phase, but it did not include structured reinforcement through 1 3F. J. Jiménez-Hornero, E. G. de Ravé Page 23 of 29 51 spaced retrieval, low-stakes testing, or repeated guided application after the course ended. From this perspective, the observed decline in performance, especially in procedural, tool- dependent skills, is not unexpected and should not be interpreted as evidence that visu- alization learning is inherently unstable. Rather, it suggests that a one-time instructional intervention is insufficient for long-term retention of complex data-visualization compe- tence. Research on retrieval practice and test-enhanced learning has shown that repeated recall opportunities, distributed over time and supported by feedback, can improve long- term retention and application of knowledge (Schwieren et al., 2017; Larsen et al., 2013). Future studies should therefore compare the present course-based model with evidence- informed alternatives that incorporate spaced practice, retrieval-based exercises, cumulative low-stakes quizzes, or follow-up graphing tasks embedded across the doctoral trajectory. Another important limitation is that the study did not directly document the learning experiences in which participants engaged between the three assessment points. Although the doctoral context is relevant for interpreting the transition from formal instruction to more independent research activity, the present design did not examine which features of the doctoral program, supervisory practices, research tasks, or informal learning opportuni- ties may have shaped retention outcomes. As a result, the study cannot determine which specific programmatic conditions supported or hindered the maintenance of data-visualiza- tion competence over time. Future research should therefore investigate these intervening experiences more directly, for example through longitudinal tracking of research activity, supervisory feedback, methodological use in ongoing projects, and students’ self-reported opportunities for continued practice. The study was conducted in a single institutional context with a modest sample size, which may limit the generalizability of results. Although linear mixed-effects modeling effectively addressed within-subject dependency and missing data, subsequent work should include larger, more diverse samples and possibly nonlinear modeling to explore alternative trajectories of change. The one-year observation period also captures only an initial phase of skill retention. Longitudinal studies extending over multiple years could determine whether procedural decline stabilizes, reverses, or continues as doctoral students advance toward dissertation work. Moreover, future research should investigate the mechanisms underlying retention and loss. Qualitative approaches, such as reflective journals, interviews, or think-aloud proto- cols, could complement quantitative models to illuminate how doctoral students perceive the relevance and use of data visualization in their ongoing research practice. Examining environmental factors such as access to visualization software, supervisory support, and par- ticipation in research communities of practice would help explain the contextual influences shaping skill maintenance. As Sinclair et al. (2013) note, doctoral learning environments are powerful mediators of skill development, and institutional culture often determines whether early gains are sustained or dissipated over time.

    Conceptual and procedural knowledge retention in data visualization: a longitudinal study of doctoral learning and instructional design implications · 2026 · DOI

  • Viscursive approaches to VIS*H evaluation address the asymmetry that visualizations are labor-intensive to create and slow to adapt compared to text-based discourse, preventing rapid critical response and iterative debate. The field needs frameworks and technological infrastructure that enable continuous, responsive critical discourse around data visualizations.

    Chasing Meaning and/or Insight? A Survey on Evaluation Practices at the Intersection of Visualization and the Humanities · 2026 · DOI

  • Critical-theoretical evaluation approaches in VIS*H remain underdeveloped. The field lacks methodologies for systematically revealing and evaluating visualizations that naturalize, mask, or distort underlying assumptions through hermeneutics of suspicion and critical-theoretical probing of tacit schemes in visualization design.

    Chasing Meaning and/or Insight? A Survey on Evaluation Practices at the Intersection of Visualization and the Humanities · 2026 · DOI

  • VIS*H visualization evaluation lacks integration of interpretivist quality criteria (informed, reflexive, abundant, plausible, resonant, transparent) into standard evaluation practice. The field needs concrete operationalization of how these criteria replace or complement performance metrics when assessing visualizations as interpretive, meaning-pursuing artifacts.

    Chasing Meaning and/or Insight? A Survey on Evaluation Practices at the Intersection of Visualization and the Humanities · 2026 · DOI

  • The 'interview paradox' in VIS*H shows 47% of papers use interviews but interview-only evaluations receive low quality scores. The field needs rigorous workflows that capture interpretive practices in situ beyond post-hoc reflection, specifically using methods like think-aloud protocols or interaction log analysis during scholarly task execution.

    Chasing Meaning and/or Insight? A Survey on Evaluation Practices at the Intersection of Visualization and the Humanities · 2026 · DOI

  • VIS*H evaluation lacks operationalized methods for validating theoretical grounding in humanities theories (Actor-Network Theory, Post-structuralism, intersectional feminism). Current practice relies on discursive validation and contestation of coherence, fruitfulness, and explanatory power, but concrete evaluation criteria and validation procedures aligned with humanities epistemology remain underdeveloped.

    Chasing Meaning and/or Insight? A Survey on Evaluation Practices at the Intersection of Visualization and the Humanities · 2026 · DOI

  • Only 19% of design study papers in the VIS*H survey engaged with uncertainty representation at all. The field lacks VIS*H-specific adaptations of accuracy-focused evaluation frameworks and methods, and needs explicit assessment of tradeoffs between added representational complexity and uncertainty modeling introduced early in the visualization design process.

    Chasing Meaning and/or Insight? A Survey on Evaluation Practices at the Intersection of Visualization and the Humanities · 2026 · DOI

  • This exploratory study focused on designers’ perceptions and ex- periences in controlled stock image selection tasks and does not capture downstream design activities such as layout composition or integration with text and branding. The participant pool was lim- ited in size and demographics, and the evaluation relied primarily on subjective and comparative measures rather than long-term or in-the-wild use. Future work should examine behavioral traces of exploration in greater depth, involve larger and more diverse popu- lations, and study how diversity control interacts with later stages of the design pipeline. In addition, given the computational cost of DPP algorithm’s sequential sampling, scaling diversity-aware exploration to larger image collections and incorporating alterna- tive feature representations remain important directions for further investigation.

    DiverXplorer: Stock Image Exploration via Diversity Adjustment for Graphic Design · 2026 · DOI

  • Through an analysis of Amazon bestsellers, a review of limited literature on cross-cultural data visualization, and an integration of research from fields such as visual communication, web design, marketing, psychology, professional communication, and user experience, this article highlights five visual design challenges for cross-cultural users: color, images, minimalism, visual hierarchy, and textual elements.

    Commentary on the Visual Design Challenges for Cross-Cultural Users of Business Data Visualizations · 2025 · DOI

  • Little is known about the extent to which interrupted time series analysis (ITSA) can be applied to short, single-case study designs and whether those applications produce results consistent with visual analysis (VA).

    Comparing Visual and Statistical Analysis in Single-Case Studies Using Published Studies · 2015 · DOI

  • (e.g., two p-views than different about 50 msec faster (p) of of different objects with the viewpoints the same name did not two were similar (e.g., of whether (e.g., a p- and an e-view). to account for all of these results simply code and a name code. At least three codes may in terms of two thus code visual in processing pictures that preserves; second, codes: a physical be needed (2-D) stimulus-description) ( i.e., an object-description) an orientation-free from information picture, of objects. This 3-D code could concern represented rather ( Sutherland, description about perspective, than stimulus 1973). in terms of surfaces and their and stored represented information features about of objects: first, a two-dimensional about orientation (3-D); and third, a name code. For information a three-dimensional (i.e., a code the 3-D code, visual of the depicted object may be produced other depth “laws” governing features relationships specific cues present in the the construction to the object with each other, regions in terms of 2-D bounded PROPERTIES OF CODING PROCESSES IN OBJECT NAMING tasks in which a required.

    The role of visual and semantic codes in object naming · 1974 · DOI

  • T E A C H I N G MACHINES A N D PROGRAMED I N S T R U C T I O N " 2 0 9 the anthropologist views him as part of a culture. The theologian focuses attention on his spiritual aspects and relates them to a pre- sumed divine scheme . . . . To the psychologist alone falls the problem of the complete psycho- physical organization. In principle he cannot be satisfied with segments of persons related to outer coordinates. He must consider the system as a whole, and show how part systems are related to one another (Becoming, Yale University Press, ~955). Perhaps we may also be able to identify some reinforcers which will work under certain conditions; but, given the infinite com- plexities of man, we know that there is no guarantee that they will work under all conditions and with all students. Today the field of programed instruction seems to have reached a plateau where all those actively involved in the field are caught up in the examination and reexamination of all that has gone before as well as the continuing task of considering all relevant conclusions and new viewpoints. The field appears to be under- going a period of intellectual incubation prior to some new ma- jor breakthroughs. The evangelistic linearists and outspoken advocates of intrinsic techniques are less vocal, and all who would program are now concerned with more complex confron- tations than in the past. What will the next so years hold for this field? An exciting possibility is the potential contribution of those who would an- alyze the total gestalt. Perhaps there will be a successful dis- covery of new reinforcers to incorporate into individualized units of instruction. A closer association with the field of basic research may be hoped for, but more difficult to obtain. A look back in so years at today's programming styles will show our techniques to be as outmoded as the Model T. This will be because of "new models" which incorporate greater use of visuals and which depart from the current concept of what con- stitutes a frame. Also there will be development and adaptation of programming techniques, or, more specifically, the philosophy of programming to motion pictures, television, and all forms of media. In addition, there must be the attempt at translating the "gains" in instructional technology developed in the laboratory environment of the military to the educational community. And finally, a pretesting and posttesting of an entire educational sys- tem may eventually materialize, with major revisions accom- plished based upon analysis of feedback. Quo vadis, programed instruction?

    A view from the terminal frame · 1964 · DOI

  • We presented our findings in the form of an integrated taxonomy, discussed how ex- isting systems populate the design space, and highlighted areas of concentration and underexplored regions.

    A Scoping Review of Mixed Initiative Visual Analytics in the Automation Renaissance · 2026 · DOI

  • Theprotocoldiscourse’suniquepropertiesrepresent the possibility of further study of the possible classification of theapplication fromwhich thenetworksession wasgenerated.

    Data Fingerprinting and Visualization for AI Enhanced Cyber-Defence System · 2026 · DOI

  • We argue that the open problem is not chart code generation but chart publication: making the output look like a top-venue figure, survive the target layout, and respond to precise author edits.

    Demonstrating chart-plot: Closing the Last Mile of Academic Chart Generation · 2026

  • Moving forward, future research should focus on refining the quantitative relationships between different types of heterogeneous traffic and extending the model to incorporate a broader array of influencing factors.

    Fusing Visual Quantified Features for Heterogeneous Traffic Flow Prediction · 2024 · DOI

  • Although multiple resources related to graph construction and visual analysis are available in the literature, studies on training pre-service and in-service professionals to visually analyze data are lacking.

    Issues and Improvements in the Visual Analysis of A-B Single-Case Graphs by Pre-Service Professionals · 2019 · DOI

  • " Until recently the computer has been limited to the analysis of numbers and words; now programs are being developed to read, analyze and display all kinds of pictorial information: maps, diagrams, photographs, even video tapes and disks.

    Computer-Assisted Mapping · 1984 · DOI

Most-cited papers in Data Visualization and Analytics

Most recent work

Find a gap in your own Data Visualization and Analytics sub-topic

This page shows what the Data Visualization and Analytics literature already flags as unresolved. To narrow it to your specific question, run the guided finder — it searches the gap library on demand and checks candidates against 250M+ OpenAlex works.

Open the Research Gap Finder →

Related topics in Computer Science

Online Learning and Analytics
363 open questions
AI in Service Interactions
278 open questions
Teaching and Learning Programming
239 open questions
Digital literacy in education
230 open questions
Mobile Learning in Education
183 open questions
Virtual Reality Applications and Impacts
178 open questions
Computational Drug Discovery Methods
126 open questions
Blockchain Technology Applications and Security
121 open questions
31 open questions have been extracted from the limitations and future-work passages of 1,192 Data Visualization and Analytics papers in our library. Each one below links back to the study that raised it, so you can read the original claim in context.

Tools for your next paper

Gaps & Literature SearchFind open questions in any field — 120,000+ open gaps mined from our own library.Pre-CheckIs your paper ready? A Tier 1–5 acceptance probability in seconds.Journal FinderA ranked shortlist of target journals from a 17,500-venue index.Duplicate Publication CheckerCheck prior-publication & salami-slicing risk across 8 sources.Citation GeneratorFreeA DOI or title → APA, MLA, Vancouver, Chicago, BibTeX, RIS. No signup.Graphical Abstract MakerFreeTurn your findings into a colorblind-safe graphical abstract.

Compare the category — Honest roundups of the AI research tools, ours listed alongside the alternatives.

Command palette

Jump anywhere, run any action.