Depicting More Information in Enriched Squarified Treemaps with Layered Glyphs

Round 1

Reviewer 1 Report

This paper presents a visual strategy combining treemaps and glyphs to represent layered/multidimensional data. The authors also perform a comparison of 3 resulting designs: treemap depicting all the layers, treemap depicting the first 3 layers with glyphs depicting the last 2, treemap depicting the first layer and glyphs depicting the others. The evaluation was made by 36 participants who had to resolve 8 tasks on one strategy.

I really appreciated the soundness of the evaluation. It is also worth mentioning that an application to create the treemaps is available, which is a very good point. For these reasons, I think the paper should be published with minor changes described hereafter.

My main concern is about the visual variables employed to validate the approach. In Figure 5 p8, an ordinal attribute (Low - Medium - High) is mapped on shapes. Please refer to Munzner’s book, Visualization Analysis and Design, to map appropriate type of attribute on the different visual variables. Figures of the book:
- Type of variables : https://www.cs.ubc.ca/~tmm/vadbook/eamonn-figs/fig2.4.pdf
- Use of visual variables : https://www.cs.ubc.ca/~tmm/vadbook/eamonn-figs/fig5.1.pdf
In the same vein, in p9, some ordinal attributes are classified as categorical, like Temperature_CAT or RainIntensity. Categorical visual variables are used to map them, while ordered visual variables should be used. I don’t think this invalidates the study, but it is clearly not optimal, and this should be discussed. In the same vein, « continuous variables » all along the paper should be avoided: the important thing in visualization is whether they are ordered (quantitative and ordinal) or not (categorical).

Missing references:

- In the introduction and in the related work on treemaps, you should mention treemaps with geographical-like shapes: D. Auber, C. Huet, A. Lambert, B. Renoust, A. Sallaberry, A. Saulnier. GosperMap: Using a Gosper Curve for Laying out Hierarchical Data. IEEE Transactions on Visualization and Computer Graphics (TVCG), 19(11): 1820-1832, 2013.
- The book of Munzner could be mentioned in section 2.1, as it is more recent and complete than the ones cited: T. Munzner. Visualization Analysis and Design. A. K. Peters, 2014.

Questions and suggestions:
- In 3.1, it is not clear why you selected [37], [38], [39], [40]. I suppose there are many other articles showing examples of glyphs? And why [39] does not appear in Table 1? Please explain.
- I suggest to add a Figure example in the introduction or at the beginning of the section 4 to ease the comprehension by giving an idea of what your are actually proposing.

Minor comments:
- p1: structures[1] -> structures [1]
- In the introduction p1, [2] is cited for circular treemaps, while [16] is cited in the related work p3. Please be consistent.
- p4: ; However -> ; however
- p4: Maguire et al.[38] -> Maguire et al. [38]
- Figure 2 p5: to A.. -> to (a).
- p7: After to complete -> After completing?
- p9: Particiapnt’s profile should be developed : how many men and women? how many graduates and undergraduates?
- Figure 8 p 11: complemente?
- p16: the use o glyphs -> the use of glyphs
- p17: 33 volunteers -> 36 volunteers?

Author Response

Reviewer: My main concern is about the visual variables employed to validate the approach. In Figure 5 p8, an ordinal attribute (Low - Medium - High) is mapped on shapes. Please refer to Munzner’s book, Visualization Analysis and Design, to map appropriate type of attribute on the different visual variables. Figures of the book:

- Type of variables: https://www.cs.ubc.ca/~tmm/vadbook/eamonn-figs/fig2.4.pdf

- Use of visual variables : https://www.cs.ubc.ca/~tmm/vadbook/eamonn-figs/fig5.1.pdf 

In the same vein, in p9, some ordinal attributes are classified as categorical, like Temperature_CAT or RainIntensity. Categorical visual variables are used to map them, while ordered visual variables should be used. I don’t think this invalidates the study, but it is clearly not optimal, and this should be discussed.

In the same vein, « continuous variables » all along the paper should be avoided: the important thing in visualization is whether they are ordered (quantitative and ordinal) or not (categorical).

The semantics over the synthetic data is just for communication purposes. When generating the data, we considered all non-quantitative variables to be categorical, so some category names indeed seem slightly off, as they imply ordinal types. We added this as a limitation in the conclusion. Additionally, we made the following steps to address this issue for the readers:

• we replaced the term continuous  for quantitative, and added the term ordinal and ordered when needed.
• we updated the Figure (now) 2 to explicitly state which variables are suited for categorical, and for ordinal. We also show that the color variable subdivides into hue and brightness accordingly to the underlying data, so it can be used to both types.
• we extended a paragraph in the conclusion to address this issue and point future research directions regarding new layers tailored specifically for categorical or ordinal data.

Reviewer:

Missing references:In the introduction and in the related work on treemaps, you should mention treemaps with geographical-like shapes: D. Auber, C. Huet, A. Lambert, B. Renoust, A. Sallaberry, A. Saulnier. GosperMap: Using a Gosper Curve for Laying out Hierarchical Data. IEEE Transactions on Visualization and Computer Graphics (TVCG), 19(11): 1820-1832, 2013.
- The book of Munzner could be mentioned in section 2.1, as it is more recent and complete than the ones cited: T. Munzner. Visualization Analysis and Design. A. K. Peters, 2014.

Suggested references added

Questions and suggestions:
- In 3.1, it is not clear why you selected [37], [38], [39], [40]. I suppose there are many other articles showing examples of glyphs?

We added a paragraph at the start of section 3.1 to explain that the chosen works are the ones which correlates the most with the proposed approach.

And why [39] does not appear in Table 1? Please explain.

We fixed the table, and now [39] should be at the correct place

I suggest to add a Figure example in the introduction or at the beginning of the section 4 to ease the comprehension by giving an idea of what your are actually proposing.

We added an image to the introduction to illustrate the squarified Treemap with layered glyphs earlier in the paper, highlighting the glyphs.

Minor comments: 
- p1: structures[1] -> structures [1]
- In the introduction p1, [2] is cited for circular treemaps, while [16] is cited in the related work p3. Please be consistent.
- p4: ; However -> ; however
- p4: Maguire et al.[38] -> Maguire et al. [38] 
- Figure 2 p5: to A.. -> to (a).
- p7: After to complete -> After completing?
- p9: Particiapnt’s profile should be developed : how many men and women? how many graduates and undergraduates?
- Figure 8 p 11: complemente?
- p16: the use o glyphs -> the use of glyphs
- p17: 33 volunteers -> 36 volunteers?

We corrected all of the minor comments.

We also proofread the paper and fix grammatical and style issues.

Reviewer 2 Report

In this research, authors proposed glyphs that use overlapped layer (i.e., Layered glyphs) as layers are composed into multidimensional glyphs with all layers being readable. The research method cites and combines two prominent visualization techniques, merging Treemap + Glyphs techniques. A prototype was developed and tested followed by user study to compare three scenarios of visual data mappings for Treemaps: only Glyphs, Glyphs and Hierarchy, and only Hierarchy. The effects were measured and evaluated in terms of task speed and accuracy. Evidently, layered glyphs resulted in an enrichment of the visualization capabilities of Treemaps without impairing the data readability by the users. This research report clearly states and presents all the major components of an extraordinary experiment. Literature review is comprehensive with current and related references. The research design is well developed and implemented. Figures are highly creative and visually effective in depicting results and other data/information. The conclusions directly relates to the results analysis. Overall, this is a comprehensive and good research report.

Author Response

Thank you! We appreciate the compliments

We addressed the comments of the other reviewer, so we are sending them here just for the record.

'''CHANGELIST OF THE OTHER REVIEWER

Reviewer: My main concern is about the visual variables employed to validate the approach. In Figure 5 p8, an ordinal attribute (Low - Medium - High) is mapped on shapes. Please refer to Munzner’s book, Visualization Analysis and Design, to map appropriate type of attribute on the different visual variables. Figures of the book:

- Type of variables: https://www.cs.ubc.ca/~tmm/vadbook/eamonn-figs/fig2.4.pdf

- Use of visual variables : https://www.cs.ubc.ca/~tmm/vadbook/eamonn-figs/fig5.1.pdf 

In the same vein, in p9, some ordinal attributes are classified as categorical, like Temperature_CAT or RainIntensity. Categorical visual variables are used to map them, while ordered visual variables should be used. I don’t think this invalidates the study, but it is clearly not optimal, and this should be discussed.

In the same vein, « continuous variables » all along the paper should be avoided: the important thing in visualization is whether they are ordered (quantitative and ordinal) or not (categorical).

The semantics over the synthetic data is just for communication purposes. When generating the data, we considered all non-quantitative variables to be categorical, so some category names indeed seem slightly off, as they imply ordinal types. We added this as a limitation in the conclusion. Additionally, we made the following steps to address this issue for the readers:

• we replaced the term continuous  for quantitative, and added the term ordinal and ordered when needed.
• we updated the Figure (now) 2 to explicitly state which variables are suited for categorical, and for ordinal. We also show that the color variable subdivides into hue and brightness accordingly to the underlying data, so it can be used to both types.
• we extended a paragraph in the conclusion to address this issue and point future research directions regarding new layers tailored specifically for categorical or ordinal data.

Reviewer:

Missing references:In the introduction and in the related work on treemaps, you should mention treemaps with geographical-like shapes: D. Auber, C. Huet, A. Lambert, B. Renoust, A. Sallaberry, A. Saulnier. GosperMap: Using a Gosper Curve for Laying out Hierarchical Data. IEEE Transactions on Visualization and Computer Graphics (TVCG), 19(11): 1820-1832, 2013.
- The book of Munzner could be mentioned in section 2.1, as it is more recent and complete than the ones cited: T. Munzner. Visualization Analysis and Design. A. K. Peters, 2014.

Suggested references added

Questions and suggestions:
- In 3.1, it is not clear why you selected [37], [38], [39], [40]. I suppose there are many other articles showing examples of glyphs?

We added a paragraph at the start of section 3.1 to explain that the chosen works are the ones which correlates the most with the proposed approach.

And why [39] does not appear in Table 1? Please explain.

We fixed the table, and now [39] should be at the correct place

I suggest to add a Figure example in the introduction or at the beginning of the section 4 to ease the comprehension by giving an idea of what your are actually proposing.

We added an image to the introduction to illustrate the squarified Treemap with layered glyphs earlier in the paper, highlighting the glyphs.

Minor comments: 
- p1: structures[1] -> structures [1]
- In the introduction p1, [2] is cited for circular treemaps, while [16] is cited in the related work p3. Please be consistent.
- p4: ; However -> ; however
- p4: Maguire et al.[38] -> Maguire et al. [38] 
- Figure 2 p5: to A.. -> to (a).
- p7: After to complete -> After completing?
- p9: Particiapnt’s profile should be developed : how many men and women? how many graduates and undergraduates?
- Figure 8 p 11: complemente?
- p16: the use o glyphs -> the use of glyphs
- p17: 33 volunteers -> 36 volunteers?

We corrected all of the minor comments.

We also proofread the paper and fix grammatical and style issues.

'''