Linear NoteBook vs pure graph approach

[Dmitry Ponyatov]

NoteBook is good model, but I see some very limitative problem: it is linear structure.

Does anybody takes search in non-linear graph information representation ?

I mean some structural document consists of isolated elements, groupped into graph structure.

Maybe it shuld be like and old Mac's HyperCard model -- lots of elements, linked by relational/ontology links,

and some sort of mapping & vizualisation rules from graph knowledge database into one or more linear printable documents.

So for example: I have large math knowledge database which I can manipulate visually: add text, formulas, schemes, plots,...

and interconnect this tiny elements with some links: hyperlink, is-a, a-kind-of, uses/used-in, partical-case,... 

And I can add some mapping rules set which maps elements from this KB into set of LaTeX markup files can be processed into for example set of textbooks for first graduate students, and interactve KhanAcademy-like internet learning system.

Is HyperGRaphDB and maybe Seco ready to build this like system ?

[Borislav Iordanov]

Hi,

On May 20, 2017, at 6:33 AM, Dmitry Ponyatov <dpon...@gmail.com> wrote:

NoteBook is good model, but I see some very limitative problem: it is linear structure.

It is rather linear indeed. Well, tree-like to be more accurate, but I see your point.

I do believe other UI approaches need to be explored. For writing code, the notebook is very good, but for integrating interactive elements with code and having a live environment that mixes the two does requires some innovation in that area. The one thing we’ve experimented with and got working successfully is a zoomable interface. The latest code does not start with that interface by default because it is unconventional and I thought it might have the effect of turning away first time users who just want to play with some dynamic languages in a JVM environment. I had made this video tutorial about the zoomable interface:

https://www.youtube.com/watch?v=k9fbOmc2tOk

(turn the sound off, there is unfortunately some noise in that video).

The zoomable interface is built with the Piccolo2d library, which is alive somewhere on github, but I don’t believe it’s actively maintained anymore. I believe Seco should probably be ported to JavaFX and somehow hope that doesn’t require a complete rewrite of all UI code that the notebook handling logic could be kept and somehow swing and JavaFX can be made to nicely coexist. But I haven’t investigated really, and I haven’t worked with JavaFX much.

Does anybody takes search in non-linear graph information representation ?

I mean some structural document consists of isolated elements, groupped into graph structure.

That is in fact how Seco stores its notebooks: there is the notion of “cells” which can be anything really, from scripts to visual interactive elements. And they participate in Cell Groups and one can have them displayed visually in various ways. The framework for how cells are visualized, how multiple views of the same cell can exist etc. is not super evolved (and perhaps it doesn’t need to be), but the foundation is there. And what you see in the above video with cells and bits and pieces of the notebook being detached comes from that. 

Maybe it shuld be like and old Mac's HyperCard model -- lots of elements, linked by relational/ontology links,

and some sort of mapping & vizualisation rules from graph knowledge database into one or more linear printable documents.

I don’t know HyperCard, but it seems like yes, that is the direction we have already taken. 

So for example: I have large math knowledge database which I can manipulate visually: add text, formulas, schemes, plots,...

and interconnect this tiny elements with some links: hyperlink, is-a, a-kind-of, uses/used-in, partical-case,... 

And I can add some mapping rules set which maps elements from this KB into set of LaTeX markup files can be processed into for example set of textbooks for first graduate students, and interactve KhanAcademy-like internet learning system.

Is HyperGRaphDB and maybe Seco ready to build this like system ?

Absolutely, I think yes! Do you already have that knowledge base in some form or you are planning on constructing it somehow? If so, how? What form? I would be psyched to get something like going!

In this context I should mention that there is some related work done by Axel Kramer (cc-ed) who is maintaining/developing a Mathematica-like system called SymJa - https://bitbucket.org/axelclk/symja_android_library/wiki/Home .I seem to remember that Axel did some work with showing LaTex  in Seco, but I might be mistaken. 

Cheers,

Boris

--
You received this message because you are subscribed to the Google Groups "Seco" group.
To unsubscribe from this group and stop receiving emails from it, send an email to scriba+un...@googlegroups.com.
To post to this group, send email to scr...@googlegroups.com.
Visit this group at https://groups.google.com/group/scriba.
For more options, visit https://groups.google.com/d/optout.