The constant growth in data generation, driven by technological advancement, highlights the need to organize information to extract relevant knowledge. In this context, visual representations emerge as effective tools to simplify this complex task. The automation of this process can be achieved through visualization recommendation systems. This work aims to improve the understanding of data visualization recommendations by synthesizing current literature to identify research gaps and outline initial requirements for developing prototypes and tools in this area. To achieve this, we conducted a systematic literature mapping followed by forward snowballing, covering the period from 2017 to 2025, through which we carefully selected and analyzed 89 papers on data visualization recommendations. We provide an overview of visualization recommendation systems, identifying employed techniques and categorizing studies based on different recommendation approaches. We also guide the selection of algorithms and methods for developing automatic and semiautomatic recommendation systems and present lessons learned and future research possibilities.
-
André Fernando Rollwagen is a professor at the Federal Institute of Education, Science and Technology Sul-rio-grande, IFSUL, Brazil. He is pursuing a doctoral degree in computer science at the Pontifical Catholic University of Rio Grande do Sul, PUCRS. His research interests include visualization and visual analytics.
-
Isabel Harb Manssour earned a Ph.D. in Computer Science from the Federal University of Rio Grande do Sul in 2002. Currently, she is an associate professor at the School of Technology at the Pontifical Catholic University of Rio Grande do Sul (PUCRS), Brazil and is a fellow of the CNPq technological productivity scholarship (level 2). A member of the Postgraduate Program in Computer Science, Manssour advises undergraduate, master's, and Ph.D. students. Besides teaching algorithms, programming, and data visualization in undergraduate and graduate courses, she develops research projects funded by agencies and companies such as CNPq, FAPERGS, HP Brazil, and Petrobras. Her research interests include all areas of Visualization, Visual Analytics, and Computer Vision.
The figure Methodology_Process.png summarizes our two-stage review workflow, which combines a Systematic Literature Mapping (SLM) with forward snowballing.
-
Systematic Literature Mapping – Planning
We first defined the research protocol: research questions, review process, search strings, target databases, and inclusion/exclusion criteria.
-
Systematic Literature Mapping – Execution
- Study identification. We applied the search strings and publication period to four digital libraries (IEEE Xplore, ACM Digital Library, ScienceDirect, and Wiley). The results were imported into the StArt tool, where duplicates were removed to obtain the initial set of unique papers.
- Study selection. Titles and abstracts were screened using exclusion and inclusion criteria. After this filtering, a smaller set of candidate papers was kept for full-text reading.
- Data extraction. The selected papers were read in full and coded according to predefined extraction criteria. The extracted data were exported to a spreadsheet for organization and synthesis.
-
Forward Snowballing – Execution
Starting from the primary studies obtained in the SLM, we performed forward snowballing in Google Scholar.
- Study identification. We retrieved all citing papers and removed duplicates.
- Study selection. Using the same inclusion and exclusion criteria, we screened titles/abstracts and then full texts, retaining additional relevant studies.
- Data extraction. These new papers were also read in full, and their data were exported and integrated into the same spreadsheet.
-
Summarization and reporting
In total, the combined process (SLM + forward snowballing) yielded 89 primary studies on data visualization recommendation. The extracted data were then synthesized to support the analyses, lessons learned, and future research directions reported in the paper.
-
GraphicalAbstract.png - Graphical Abstract.
-
Methodology_Process.png – Diagram of the research methodological process.
-
Overview of studies - technologies and algorithms - The electronic spreadsheet contains a categorization of studies organized by:
study title; publication year; uses of machine learning, uses of deep learning; uses LLM; LLM methods used; LLM methods or techniques applied; methods and algorithms used; technologies used; uses ranking in recommendation; visualization techniques used; type of contribution - (1: tool development for visualization recommendation, 2: evaluative framework development, 3: design of new methodologies, or 4: evaluation of tools); recommendation approach (knowledge-based, data-driven, user preferences oriented, task-based, or other approaches); recommendation method (automatic or semiautomatic); perceptal feature used; complex visual representations used; User Interface / API used; forms of interactivity; implements visualization tasks; implements prototype, tool, or system; evaluation (quantitative / qualitative).
- Completeness of documentation – The electronic spreadsheet summarizes, for each study, the availability and completeness of artefacts required for reproducibility. It is organized by:
study ID; study title; reproducibility flag (overall judgement of whether independent replication seems feasible); availability of source code (y = yes, n = no, p = partial); availability of dataset documentation and access information (y/n/p); description of evaluation resources (e.g., experimental design, metrics, user studies) (y/n/p); availability of supplementary material (e.g., OSF packages, demos, tutorials, configuration files) (y/n/p); ordinal documentation score “Documentation (1 = partial; 2 = adequate)”; and a qualitative observation field that briefly describes strengths and gaps in the documentation, artefact availability, and practical reproducibility for each study.