Deconstructing Categorization in Visualization Recommendation: A Taxonomy and Comparative Study

Visual Arrangements of Bar Charts Influence Comparisons in Viewer Takeaways

“…Visual analysis tools, such as visualization recommendation (VizRec) systems, can help people gain insights quickly by providing reasonable visualizations. While a detailed review of visualization recommendation (VizRec) systems and techniques is beyond the scope of this paper, it can be found in survey manuscripts such as [10,41,69,75]. Broadly speaking, VizRec systems can be classified based on whether they suggest visual encodings (i.e., encoding recommenders) [44,45] or aspects of the data to visualize (i.e., data-based recommenders) [68].…”

MEDLEY: Intent-based Recommendations to Support Dashboard Composition

“…There is a plethora of research on visualization recommendation techniques and systems, a detailed review of which can be found in other survey manuscripts [9,22,53,55]. The key idea of our work, however, is the notion of a dashboard's analytic intent, which is similar to the concept of visual analysis tasks proposed in prior literature (e.g., [4,6,35,37]).…”

“…Similar to Foresight, systems like Voder [41] and Datasite [11] also recommend visualizations but encompass them within textual data facts/insights (e.g., "Displacement and Horsepower have a strong correlation") that, in turn, correspond to analytic tasks (e.g., correlation). Frontier [22] is another example of a system that categorizes its recommended visualizations within different tasks (e.g., flter, distribution, correlation) to guide data exploration. User studies of these systems have shown that using tasks or intents as an organizing principle facilitates various data exploration strategies while also aiding interpretation of the recommendations themselves.…”

MEDLEY: Intent-based Recommendations to Support Dashboard Composition

“…Similar to Foresight, systems like Voder [41] and Datasite [11] also recommend visualizations but encompass them within textual data facts/insights (e.g., "Displacement and Horsepower have a strong correlation") that, in turn, correspond to analytic tasks (e.g., correlation). Frontier [22] is another example of a system that categorizes its recommended visualizations within different tasks (e.g., filter, distribution, correlation) to guide data exploration. User studies of these systems have shown that using tasks or intents as an organizing principle facilitates various data exploration strategies while also aiding interpretation of the recommendations themselves.…”

“…Combining recommendations with manual view specification. Prior work on visualization recommendations has demonstrated the value in combining manual view specification (MVS) and systemgenerated recommendations (e.g., [11,22,41,52,54]). MEDLEY allows users to right-click on views in the canvas and change options like the sort order, axes swapping, or data aggregation.…”