Welcomes ideas for how visually display the relationships between MANY tiddler

[magev958]

I have created a private wiki about my orchids. Each genus has got its own tiddler that links to each species within the genus.

Now I would like to show the relationship between the subfamily (5) -> tribus (22) -> subtribus (58) -> genus (769) -> species (27863) but I do not really know, without it gets far too big.

I would like to have some ideas about how and what it could be done. Any idea, no matter how crazy :)

/Magnus

[@TiddlyTweeter]

Ciao magev

As far as I understand botany species and sub-species can be represented via a tree structure? That could work in TW visually as there are several visual tools for that. I'm just not sure if plant hybrids fit that neatness. IF it sounds like a simple branching visual way to depict hierarchy could work let me know and I'll dig out some examples.

Best wishes
Josiah

[TonyM]

Magev,

What exactly is the relationship you want to highlight? They should already be in a hierarchy in your wiki if you did it correctly.

I presume you have children tagging parents and looked at the TagglyTagging for TW5 of Tobias (great for automatic hierarchies based on the current tiddler), you could also use the count widget to provide totals of the children or siblings of every tiddler (put it in the viewteplate).

If the issue is your hierarchy has too many children in each level you can do at least the following;

Establish additional categories and divide each group into subgroups at each level, alternatively if this would be too much work and not applicable just divide a list of siblings alphabetically with a first letter, or in the case of naming conventions there are often classes prefixes in the name.

You could set a field for your grouping prefix in each tiddler, or you could use Prefix[A] in a filter to list all beginning with A

Ask again if you cant work it out.

Tony

[PMario]

On Wednesday, November 8, 2017 at 8:50:43 PM UTC+1, magev958 wrote:

Now I would like to show the relationship between the subfamily (5) -> tribus (22) -> subtribus (58) -> genus (769) -> species (27863)

Very interesting. ... So does it mean that your tree has ~29000 elements, or 769 x 28000 -> ~21,000,000 ++ ?

How is the naming convention?

How does your query language look like?

eg: is it needed, that you calculate a "nearest path" between 2 elements in your database?

have fun!

mario

[AndrewMc]

It will be ~29000 elements (or nodes or vertices, depending on how you refer to the parts of tree structures).

[PMario]

On Wednesday, November 8, 2017 at 8:50:43 PM UTC+1, magev958 wrote:

Now I would like to show the relationship between the subfamily (5) -> tribus (22) -> subtribus (58) -> genus (769) -> species (27863) but I do not really know, without it gets far too big.

I don't know botanic names, but is there a lot of repetition in the geneus or species naming. If there is... it may be possible to compress this stuff quite a bit.

Do you have some examples? Even better would be some links ;)

In 2011, there was a very interesting discussion going on, about compressing and looking up alphabetically sorted words. ... At the end was a library, that could compress a 600kByte file into 170k and still have an incredibly fast lookup method.

doing some maths .... 28000 species x 30 char / per species is about 840k character. Which may be split into much less then 28000 tiddlers.

So in sum there may be 2000 tiddlers, where about 1000 will contain some fancy lookup tables. ...  BUT I'm sure, there are a lot of gotchas along the way :))

have fun!

mario

[magev958]

I've tried toc as a tree, but it's getting too long with almost 800 genus.

I would like to illustrate in a compact way how many genera / species are known to science right now and how many species each genus contains in relation to each other. Some genera contain only one species, other thousands.

I have an old version online. In Swedish, unfortunately, but it gives an idea. The toc to the right is hidden until hovering

http://magev958.github.io/Orchidelirium

[TonyM]

Magnus,

I was inspired by your data challenge and had a thought. I recently did a introduction to Big Data course. One of the first examples they give is processing unstructured text into word counts, the big data processing feature is to delegate this to multiple processes for fast processing then bring the result together at the end.

If you do not already have a way to divide a large list of siblings, and my Alphabetic prefix is not enough, you could run all these items through a BigData Analysis, or even feed it into a Machine learning process with the result providing some new ways to organise such lists.

Give the nomenclature of scientific classifications, I thought you could take each name and remove the last character, 2nd last, third last.... until no characters remain, You could also do the reverse, removing the 1st, 2nd 3rd until there is not more letters. now feed this new  list of gradually truncated words into a new sample and run a word count again. Using word length length to lift the priority of a given word segment.

From this you could see the word suffixes and prefixes that are the longest and include the greater number of names in a set of words including those prefixes and Suffixes. You can find sets that only have none, one or a few members, and other sets that have many members.

Then you could apply the top prefixes or suffixes to see how often they occur under the structured  subfamily (5) -> tribus (22) -> subtribus (58) -> genus (769) -> species (27863)  hierarchy.

I expect the result will provide a whole new layer of organisation you can impose on a large set of siblings.

You many even gain new insights into how the nomenclature has developed.

Perhaps, I don't understand the nature of your challenge enough, but this may be food for thought.

Regards
Tony

[@TiddlyTweeter]

Ciao magev

I'm replying here not so much from a TW perspective as a basic organizational one.

To generate a methodology for being able to navigate a decision tree on plant species I recommend looking at the way botanists have dealt with this issue. After all its really a botanical issue. There is a long history in botany that is pertinent. For instance, if you take the genus Rosaceae, renowned for it mutable cross-pollinations, HOW on earth can you cope with that complexity? Botany developed methods for doing so.

Adapting / studying how botanists think may help you?

I think the issue is as much a conceptual issue as an issue about coding.

Best wishes
Josiah

magev958 wrote:

I've tried toc as a tree, but it's getting too long with almost 800 genus.

http://magev958.github.io/Orchidelirium

[@TiddlyTweeter]

Ciao magev

Part of the issue is branching overload.

For this kind of thing "working backwards" may help. By which I mean can you visualise what the tree should look like at its branchings to be manageable? (E.g. 600 items listed in a Tiddler is likely too much, so WHAT is the SUB-division? 10, 50, 100?).

ONE suggestion is to add further levels so that, for instance, genus, is sub-divided in some way ... maybe by morphological features of the plants by genus groups (if possible), or by simple grouped indices on names.

In your current thinking the hierarchy is a strict descent line. But there is no reason I can see you could not interlace descriptive or denotative groupings that could make the volume more manageable in its presentation. That's what botanists tend to do in practice in complex situations. The classic "botanical key characteristic identification system" has a lot of utility I think.

Just a thought from an amateur botanist :-).

Best wishes
Josiah

[coda coder]

On Friday, November 10, 2017 at 3:40:14 AM UTC-6, @TiddlyTweeter wrote:

I think the issue is as much a conceptual issue as an issue about coding.

 

In gold, like it should be.  (Can't believe you're not a programmer, sometimes.)  Errors at this key stage lead to the infamous "form over function" mistakes in most bad designs (and not just in programming).

[PMario]

Hi,

On Friday, November 10, 2017 at 2:52:33 AM UTC+1, magev958 wrote:

I've tried toc as a tree, but it's getting too long with almost 800 genus.

I'm not at the stage, yet, where I think about the visual representation. ... I was thinking about the possibilities how to store the info, without creating tiddlers...

I would like to illustrate in a compact way how many genera / species are known to science right now and how many species each genus contains in relation to each other. Some genera contain only one species, other thousands.

I did find a very interesting page: https://tree.opentreeoflife.org/opentree/argus/opentree9.1@ott568878/Orchidaceae  ... Would this visual design be closer to your needs?

have fun!
mario