I am still working on how to best show citation sources in the Dynamic View.
To show them by full name will take too much space, considering an average PhD contains (anecdotally) around 200 citations and showing dot icons do not provide much information. There are of course well-established interactions to deal with with this, including mouse overs and lensing views and the view will need to accommodate extra nodes including headings, annotations and possibly ordering elements such as lines and boxes.
User can drag citations onto the dynamic view from a valid PDF in a folder. On import they will be checked for Visual-Meta and if none are available they will automatically be sent to Scholarcy for processing and appending Visual-Meta to the document. What appears in the Dynamic View are essentially the citation/BibTeX/Visual-Meta for the document–the document itself is left where it was.
Screen real estate experiments
At the back of this document I laid out some of my own literature review documents and it became quickly unreadable, with something around 100 items:
Cutting the titles down to first part (if a punctuation divided the title) and all capitalised the same, it looks like this:
I am using a 15” screen for this work and these are screenshots from full screen. I can scale the text down as much as shown here, to about 75%, and it is still readable:
I further abbreviated the text to similar lengths with ‘…’ ellipsis for more uniform sized-nodes and the reduction is to 65% and the area used (Photoshop manipulation). In this version the headings are also small:
But here the headings are larger (non-scalable or separately scalable):
In can see how pinch to scale the whole or selected documents or even select a category such as a document or a heading and pinch to scale all of the same size and maybe spacebar to make selected documents open and larger, such as also showing authors. The point is: The interactions need to be explored for scaling.
Interactions for Screen Optimisation
Connections to headings should not require visual lines but connections between citations should be visible. To connect citations to a heading select citations and drag and drop onto a heading. Animation then flashes heading and scales, and citations moves to original location. Ctrl-drag a node onto another and a line appears.
User can select one or more headings and ‘spacebar’ to snap any connected citations to it. No visible change to the heading.
Annotations should be differently scaled to citations and headings somehow.
Further Questions About Interactivity
Would it be useful to have a notion of infinite canvas or have a storage place outside of the canvas? Perhaps ‘inside’ the toolbar on the bottom?
How can the user change how much text is shown for each node? Drag border or click and -/+?
And the most pressing question remains; how useful is this for actually showing a teacher that I, as a PhD student, have done the LR to a sufficiently deep and broad degree?
I think that’s the wrong question at this point. I think the issue is how to allow a reader, teacher or otherwise, to interrogate the citations in a published document to see how the document is connected to its environment?