Ploceus

I had the opportunity to try out Ploceus 0.1 by Zhicheng "Leo" Liu at Georgia Tech. This page contains notes that I've taken while working through the tutorial PDF.

In his words,

Ploceus supports flexible modeling of tabular data as graphs or 
networks as well as visual analysis of the modeled networks.

The PDF tutorial is very well organized and makes it easy to orient when getting started.

Some silent hiccups when getting started:

/Users/me/Desktop/ploceus/version/201: command not found/../etc/ploceus.conf: line 4: 
/Users/me/Desktop/ploceus/version/201: command not found/../etc/ploceus.conf: line 10: 
/Users/me/Desktop/ploceus/version/201: command not found/../etc/ploceus.conf: line 13:

Straightforward to start on the Mac (double-click). An icon would help indicate that ploceus/bin/ploceus should be double clicked (also, this is slightly different than what is stated in the tutorial: ploceus.app). Very straightforward to increase memory allocation, according to the tutorial.

Red labels in Figure 1 are very useful. Would be more useful if the labels were cited in the narrative as they are described (e.g. "The data table column view (a)").

I'm wondering what mapping language is being used underneath (and what the resulting "network schema" involves at a technical level).

I'm concerned about the disconnect between the data table and the data sets panes. Is this intuitive for someone unfamiliar with the data? I would think that these two views could be unified in some natural way.

Section 4.0 is a very informative introduction.

I'm a bit confused about what a "tab" shows. Is it a new data transformation (table->graph) or a new view on a same data transformation (or both, and how could you tell...)?

I'll try using the spreadsheet of US Farmers Markets from data.gov.

Seems a bit odd that "4.1 Managing Networks" is discussed before we even load data. I guess I think of my source data as "non network", and I load that in and THEN make a network. Do I need to "make a network" in the application and THEN load up some tables?

By the time I get to Figure 2, I see that some of the UI icons are semiotic in nature. For example, the green plus or red x are composed with the yellow graph. Red x and "table" icon is delete a table. What is script G and P? Hover overs indicate "aggregate" and "project", so the app is helping me decode the icons (yeah!). So is "yellow" data and "blue" view? The aggregate icon's lime background is a grouping, so that makes sense. The red portions of "project" are distinguishing themselves from orange portions. Makes sense. Thanks!

In the tutorial, before taking us into using any of the buttons, perhaps a table of the (isolated) icons and showing their hover text would help give an overview before we take some detail. Adding a column in this table showing how to get the equivalent action from the menu would also be very helpful. Using slash notation to indicate menu hierarchy is a bit confusing. For example, Edit/New Network seems to indicate that one is either Editing OR "New"ing a network. I'd recommend using a notation such as Edit : New Network which is much more common in tutorials. Also, putting it into a different font would help, instead of wrapping it in quotes.

Why are we creating a network when there is already a network created when we load the application? It's sort of distracting and crufty to have the original one hanging around.

5.1 "To import these sample data sets, simply click on the “Data/Build Sample Data” menu item" - it would be helpful to summarize what "build" means in this situation, as this is not a conventional term. What is involved?

Selecting a sheet when loading a spreadsheet is very responsive. Nice!

The up and down arrows to change the header row seem backwards. Also, might be nice to just "double click" on a header row.

Figure 7 could benefit from before and after subfigures. One for the start of the drag and one for where it is dropped. Perhaps a diagram illustrating that. "Load Table" glyph does not appear on my Mac when I grab the table name from the Data Sets pane.

6.1 "In the Ploceus interface, you create nodes by dragging selected table columns in the data management view" Nothing the the UI says "data management view" - "Data Sets" or "Data Tables"? I'd recommend referencing the panes more formally. For example, instead of "network schema view", use Network Schema View in a different font so that it is easier to recognize the views in the narrative. Also, be consistent in how you name the panes.

Reordering columns in the "Data Tables" view does not change the order in the "Data Sets" view.

I closed all of my "Network tabs", and when dragging "LastName" and "FirstName" to the Network Schema pane, nothing happened. Please indicate that no Network exists and ask me to make one (or just create one for me).

Once dragging the "LastName" and "FirstName" columns to Network Schema view, I wanted to remove them but couldn't right click or undo.

When making a "LastName, FirstName" node, I saw the gridded nodes in the visualization. After zooming in and out a few times, they disappeared. Did I zoom out too far? I can't seem to get them back.

It was easy to make a network schema between "LastName, FirstName" and "MeetingRm" nodes, and it was obvious that something interesting was being shown, but I don't know how to interpret the graph.

Adding "visitor count" to the "LastName, FirstName" node seemed to make the nodes different sizes. Surprising, but nice, that the view found a quantity to use in the display.

The graph layout does not avoid overlaps as much as I'd like them to. I get distracted by them.

Looks like you cleaned the data, no "Visitor tour" spam like in the original dataset. So, the tool can't help me recognize this noise and remove it?

Checkbox in view tab says "Node L.." regardless of how big the window is and does not provide a hover text.

The view often gets uncentered and there isn't a clear way to center it (I'd rather just be able to drag the canvas).

Edge label "John Smith - Residence - 19" - does this say that "LastName, FirstName" visited the Residence 19 times (and, there are 19 rows with these values)?

I made a new graph "visiteeLast, visiteeFirst" and "LastName, Firstname, MidName" and the two colors are very hard to distinguish.