CS-5630/6630 | Visualization | Fall 2014
Week 15: The Flavor Connection - Printable Version

+- CS-5630/6630 | Visualization | Fall 2014 (http://www.eng.utah.edu/~abigelow/bulletinBoard)
+-- Forum: CS-5630/6630 | Visualization | Fall 2014 (/forumdisplay.php?fid=1)
+--- Forum: Design Critiques (/forumdisplay.php?fid=3)
+--- Thread: Week 15: The Flavor Connection (/showthread.php?tid=111)

Pages: 1 2

Week 15: The Flavor Connection - mcarter - 12-05-2014 01:08 AM

I am critiquing an infographic from Scientific American's website. The interactive page can be found here: http://www.scientificamerican.com/article/flavor-connection-taste-map-interactive/

Here's a screenshot.

This visualization depicts the relationship between various recipe ingredients based on the chemical compounds dictating their flavors.

Each ingredient is encoded as a blue circle, whose area indicates the number of recipes (among a database of over 56,000) the ingredient is found in. A label is drawn for each ingredient with sufficient area, or sufficient room for display. The blue circles are categorized horizontally into different food groups, and the food groups sorted by group name. The vertical position of the blue circles encodes the number of other foods with which the indicated ingredient shares "flavors", or chemical compounds contributing to it's flavor.

A line (edge) is used to encode two foods have common flavors, with the thickness of the line encoding the number of flavors in common. Red lines encode a relationship between ingredients from different food groups, and a grey line encodes relationship between ingredients in the same food group.

The visualization is also interactive. On the left hand side of the visualization is a filter of sorts, which lets the user select a row, effectively selecting all ingredients who have relationships with the number of other foods indicated by the selected row. Selecting the 10 row removes all foods other than those who have relationships with 10 other foods. Rows with orange numbers indicate that there are foods in that row, while grey indicates there are no foods with that number of relationships.

Finally, there is also an axis on the right side, indicating the number of recipes each ingredient was found in, among the database of over 56,000 recipes. This axis is only populated when a filter row is selected, as the total number of ingredients would not fit all at once on the axis.

I find this visualization to be quite visually appealing, though a little busy. The color scheme is appealing, as well as the spatial layout. I also like the axes on the left and the right that give more precise data about the current selection.

However, this is about where the good part ends. I think that this visualization struggles to achieve any tasks. It certainly isn't useful as an analysis tool - the difference in line weights indicating number of shared compounds, as well as the area of each blue circle, are so difficult to distinguish that no actual data can be discerned.

Also, it's not particularly useful as an exploration tool. There is no way to filter by food group, so you can't easily tell if certain food groups pair well with other food groups (based on the number of foods between the groups that have relationships).

Finally, it isn't even useful as a means for presenting results, as one might expect the purpose of the visualization to be, seeing as how it is created for a periodical with a general audience in mind. For example, there is no way to easily locate your favorite foods, and see what flavors they have in common with others. Further, most of the foods lie within the bottom few rows, so a majority of the foods are occluded when no row is selected (and even when any of the bottom few rows are selected).

Perhaps the worst thing about the graphic is that it doesn't really accomplish any objective. It doesn't suit any of the tasks mentioned above, and it doesn't tell any sort of story or have a take away message. After playing around with it for a few minutes, I didn't come away having noticed any interesting trends or pairings.

In summary, I felt it was a fairly appealing graphic that really struggles to achieve a purpose.

RE: Week 15: The Flavor Connection - MukteshKhole - 12-05-2014 09:18 AM

This visualization is trying to show a lot of information. It is interesting to see how different food items from different categories are related by a flavor connection. The graph is a bit cluttered with information but the good thing is we can interactively select a food item based on the frequency with which it appears in global recipe database this clears the graph a bit.
The website provides a lengthy legend (an entire page of instructions) on how to read the graph, I guess this defeats the purpose of visualization. Secondly, I dont think that using size to encode popularity and thickness to encode how strong a relationship is a very good idea for a graph which is heavily cluttered with information. The bottom part of the graph has a lot of overlapped information and is very difficult to understand.
Overall this graph provides a lot information but one has to really put in efforts to understand it.

BTW, I think we have seen this visualization before.

RE: Week 15: The Flavor Connection - shaoyuspace - 12-05-2014 02:41 PM

For the visulization, i think it is not clear for the inforamtion. i look it and i feel to hard understand it. it has a lot of overlapped information on the margin. But i find it still have good interaction for them. and i strong argee you said it isn't even useful as a means for presenting results, as one might expect the purpose of the visualization to be, seeing as how it is created for a periodical with a general audience in mind.

RE: Week 15: The Flavor Connection - Aravind - 12-05-2014 07:41 PM

This visualization is so popular that it was done in week 11 as well.

I liked your critique, especially the below point was not brought up before
"This axis is only populated when a filter row is selected, as the total number of ingredients would not fit all at once on the axis."
From my previous comments, the visualization looks neat and complex over all but it was hard for me to get my relations between items from inter-crossing lines.
I felt that when there are overlapping blue circles, I was not able to click and select a specific food item to view related items.
I was not able to compare between the thickness of the red lines connecting the food item. I would say this size encoding could have been replaced with colors that represent a range of number of shared compounds like 0-50 in yellow,red etc. An interactive legend would be useful for this purpose as well.

RE: Week 15: The Flavor Connection - sah403 - 12-05-2014 11:40 PM

Given the data set, I cant think of any other technique that would visualize the data in a better way.

Data Highlight enables the user to see only the data of interest. However the interactivity could be better. All the items at the same vertical position are selected at once and there is no way to choose just one item. It is good that spatial position was used as a channel to differentiate food items based on the number of other items that it shares common flavors with

The visualization was made targeting chefs so that items that share common flavors could be substituted. But I doubt if the visualization would really serve the purpose, For example, coffee shows strong correlation with beer and roasted peanuts. And I bet these could be never substituted by others.

RE: Week 15: The Flavor Connection - u0869331 - 12-06-2014 09:39 AM

I agree with the critic. This visualization is visually appealing but allows little scope for comparing between the data points. Neither of the metrics, the size of the blue circles of the width of the lines connecting them. Also, to gather any information for this viz, there is a lot of scrolling up and down involved since it doesnt fit in a single page.

The feature that I do like about this is that all the interactive features are labelled well, be it highlighting a specific section of the chart of clearing all selections. There is also a help page available.

RE: Week 15: The Flavor Connection - hehe - 12-06-2014 06:01 PM

This is a cool visualization although user needs some time to get familiar with it. I agree with mcarter for the most part. The good part in this visualization is interactivity. It looks messy at first glance but you may get clear information with interactive panel. It provides filter function which only shows information that you want to see. It is using size to show popularity that is OK when having enough spaces. However, when comes to more information with limit spaces these blue circles may overlap and couldn't effectively express detailed information. In addition, this visualization uses node-link to present relationship between two items, and it draws a blue node in the center of blue circle when selecting a row. I'd like to draw nodes with a smaller circle because it is confusing people when blue circle of popularity size equals to node circle. Overall, this visualization is interesting and attractive.

RE: Week 15: The Flavor Connection - Khalid - 12-07-2014 12:50 AM

I found this visualization very informative, I just want to comment on the size of the dots which shows the frequency with which the food appears in a recipe, but as we know size is not an accurate encoding of count.

Also a line connecting two dots means that the two foods share at least one flavor, the more flavors they share, the thicker the line, which is again not an accurate encoding of numbers.

RE: Week 15: The Flavor Connection - Chaomeng - 12-07-2014 03:08 AM

This visualization is interesting at first glance, it list all the food indigence and show the relationship between food indigence that share same favors. But after playing around with it, I agree that it fails to give too much useful information about the data.
First, circles and lines are too dense at the bottom of the graph and it is hard to find one specific food indigence circle among it. It is also hard to compare the food relationship number just by checking the connecting line width.
Second, I don’t think the left side interaction that electing all ingredients who have relationships with the number of other foods indicated by the selected row, would give us too much information. Just by selecting these food indigence does not tell us the inner relationship between food and indigence group, and I do not understand why it should do that.

RE: Week 15: The Flavor Connection - u0866707 - 12-07-2014 04:13 PM

Very cool visualization especially when click the items and connection lines. The interaction is fast and the color is bright and clear. We can easily find the relation of different flavor. But I still have some parts that I don't like. For example, the pure flavors, which has a low number of flavors, are overlapped too much to see the information. And when open the website, there is even a tutorial popping out to teach people how to read the data. Since the data is not that big and complex to read through, this is not necessary. Even if the data is so complex, the designer should try to show them as easy as possible. The public reader should not try to learn a new way to read this usual data.