FrontPage SiteMap RecentChanges RecentNearChanges HowTo Blog RSS

WikiAndConnectivity

HolisticProgramming → this


[fr]Le wikiwiki (qui est déjà à l’heure actuelle un artefact dépassé) n’a d’intérêt, me semble-t-il, que dans ce qui constitue sa spécificité. En dehors de tout usage dit “social” (i.e. écriture collaborative), il ne reste qu’une seule spécificité, qui est sa gestion particulière de l’hypertexte.
[en]Wikiwiki (which is already obsolete here by now) is interesting only in those things which make it specific (methinks). The social uses put apart (that is, collaborative writing), there is only one aspect that remains : its particular way of dealing with hypertexts.

[fr]Cette dernière est en fait une émanation de l’idée de “réseau”, contemporaine du primat de l’information. Pour faire simple : connecter des données (le plus souvent des pages, et parfois des “pajobjets”) sous forme de graphe directionnel.
[en]This latter aspect is indeed a consequence of the concept of “net”, which is contemporaneous with the “primat de l’information” (~ leadership of information). To say it briefly : we connect data under the shape of a directed graph. (Most often, these data are pages, sometimes they are OOP objects (e.g. the “pajobjets” on BlueMoon))

[fr]Le côté sympa du MotWiki là-dedans est tout à fait anecdotique, et camoufle d’ailleurs la triplicité de sa nature (j’ai promis des éclaircissements à C. Nepote à ce sujet, cela viendra).
[en]WikiWord+s seem really cool, but this reputation is ill-begotten, as it is purely anecdotic. It tends to shade the triple essence of the WikiWord (I promised to give some explanations of this latter statement to Charles Nepote, and I will do it in due time (i.e. when I have some time))

[fr]Tout cela pour dire que si l’on cherche à “organiser” des données sur un wiki ou quoi que ce soit du même genre (i.e. quelque chose qui s’appuie sur la langue naturelle et non pas sur un codage artificiel, et qui d’autre part repose sur la forme du graphe orienté), alors il est naturel de chercher à exploiter les propriétés de la forme “langage naturel” et celles de la forme “graphe orienté”. Ne pas exploiter ces propriétés, c’est sous-employer l’outil qu’on utilise, ou en tout cas passer à côté de beaucoup de choses.
[en]All this was intended to mean that, if one wanna “organize” things on a wiki [or on whatever system which is the same kind as a wiki, that is : any system which is based on both 1° the use of natural language (vs. any predetermined formalism) and 2° directed graph (vs. topic trees, or the like)…], then it is natural for this person to use (and abuse) the properties of the two forms “natural language” and “directed graph”. Not to use these properties, that would be under-use the wiki tool, or at least spoil a lot of fun.

[fr]Maintenant, pour ce qui est des propriétés de la langue naturelle, celle qui se distingue le plus du caractère logiciste ambiant est le “flou” de ses déterminations, le caractère “poétique” de la langue, si l’on veut. Ce flou est même celui qu’on utilise lorsqu’on fait une recherche google (car google ne sémantise pas).
[en]Now, concerning the interesting properties of the natural language, the most specific of all is the way natural language avoids the traps in which all “logicist” formalisms fall. That is the “fuzzy” or “poetic” use of the language, if you want it put this way. This fuzzyness, this poetry are even used by us humans when we google something (for google do no semantics).

[fr]Pour ce qui est du graphe, la propriété la plus spécifique, qui le distingue de l’arbre, est la présence de cycles (un arbre est mathématiquement défini comme un graphe exempt de cycles).
[en]As for the directed graphs, the most specific property -that is, the property that distinguigh them from trees- is the presence of cycles. Remember : in mathematics, a tree is defined as a graph that has no cycle in it.

[fr]Il découle donc de ce qui précède que la meilleure utilisation possible d’un outil de type wiki utilise le caractère poétique du langage et la teneur cyclique des graphes.
[en]From what has been said, it follows that the best use of any wiki-like tool is a use in which the poetic character of language and the cyclic ingredient of graphs are used…

[fr]Ces deux déterminations font référence à ce que l’on appelle “connectivité”. On tient compte de la connectivité en faisant agir sur des structures de graphes des algorithmes de type hydromimétique qui permettent de paralléliser la computation, et de tirer du réseau sa “substantifycque moelle”.
[en]Both these two properties are the determinations of the so-called “connectivity”. They are accounted by running hydromimetic algorithms within graph data-structures, so as to parallelize computation, and get out of the net its “substantifyque moelle” (*).

[en][(*) this is a pun and allusion to Rabelais. I do not know the required translation.]

[fr]Pour information, on trouvera sur n’importe quelle page de BlueMoon un lien tout en bas de la page qui s’appelle “Clash!” et qui permet de calculer des “itinéraires de plus grandes cohérence” entre n’importe quel lot de pages ou de thèmes donnés par avance. – NicolasMontessuit aka esc
[en]For information, one will find on any page of BlueMoon a link called “Clash!” (at the bottom). This link computes the “greatest coherence path” between any number of pages or topics that one give it as an input. – NicolasMontessuit aka esc

[en]Applied multilingual formatting. Use the filters below to read one language only. MattisManzel

[en] i don’t know exactly, what you mean with “connectivity”. but as far as i can see, this is the “moelle” of a knot (node) in the XanaduVersion. the triple essence of a knot is, that it is the “connectivity” of two other datas, which can be knots too. therefore i say tree, but in fact it’s a net, because there can be the same knot in several connections, or even a knot may contain itself in its branches (recursion (but then you need a condidition for breaking off)). since there are pointers back to the roots, there are cycles too. so what? why all this polemizing? --sigi

Let I give an example, that will show how one can shift gradually from a fully logicist point of view to an holistic one. Suppose I want to represent the concept of a knife. For instance, I am a semantic-web hacker, and I have to record the existence of some knives in my server. The first idea, which is 100% logicist, is to design a TREE of all existing objects, like this :

Material-Objects ::= Sharp-Objects, Smooth-Objects

Sharp-Objects ::= Harmful-Objects, Harmless-Objects

Harmful-Objects ::= Spears, Knives, Kriss, usw.

Now, this is not so good, because one could also say that a knife is a kitchen utensil, and put a tree as follows :

Material-Objects ::= Useful-Objects, Useless-Objects, Objects-Sometimes-Useful-And-Sometimes-Not

Useful-Objects ::= Tools, etc.

Tools ::= Kitchen-Utensils, Car-Tools, Internet-Tools

Kitchen-Utensils ::= Fork, Knife, Spoon, etc.

So you see that any conceptual tree is arbitrary. And also the ‘MAIN’ category (the one at the top, which is the nearest to God) is problematic. So the idea would be to replace TREES with FIBRAGES.

That is, you superimpose each possible tree, to get another structure which condense all the trees.

The bad thing is that TREES were easy to deal with, because of the uniqueness in the process of “inheritance”. With FIBRAGES, you have multiple inheritance, which is a nightmare.

And not only that, but FIBRAGES are still directed, in the sense that the only relation between two nodes is vertical. But in real life, human beings do not only categorize with respect to verticality, they also categorize horizontally, that is, they say two things are like “brothers” or “friends” or “brides”, not only like father and son.

A question would then be : is it possible to render the horizontalness of relations with only vertical relations ? That is, say for example BROTHER(X,Y) iff FATHER(Z,X) and FATHER(Z,Y) ?

The answer is no, I do not have the room here to explain why, but believe me.

So, you have to supersede FIBRAGES with something more complex. Then the idea is to use GRAPHS. A priori GRAPHS are a good idea, because they denote horizontal relations. Wikiwikis for example are an example of a structure that is 90% GRAPH and 10% TREE… roughly speaking. But now the question is : how to exploit these relations.

Indeed, they is such a fuss of relations in every direction in a graph, that you need to get a good way of collecting data not to be lost. There are two ways of looking at the things : locally and globally.

Globally, you look at the entire structure of the graph.

Locally, you extract data from one node and its neighbours.

From what you said in response to this page, it is clear that you are in a GRAPH paradigm. This is not surprising, btw, since wikiwikis are developped on the graph “philosophy”…

GRAPHS permit us to play with “greatest coherence paths”. Suppose you have a graph where you recorded the following facts, in the form of nodes and vertices :

Ill ——– to Heal

to Heal ——- to Mend

Bugged —– to Mend

Bugged ——- a Robot

Somebody ——- Ill

to Heal —– Hospital

and other links… Then, suppose you light the node “Hospital” and the node “a Robot”, and then that you excite each neighbour of a site already lit, but with a decreasing stimulus. Once this is done, you start from one the sites you excited first, then follow the “greatest line”, that is, go on the most excited neighbour on a trip.

Assuming there are other nodes and other links but that they do not constitute a denser kernel, you will obtain : Hospital — to Heal — to Mend — a Robot, that is, your GRAPH structure will have returned you the MEANING of the METAPHORIC EXPRESSION “the hospital of a robot”.

But then, if you scrutinize it, you will see that the structure which is used is a little complex than a simple GRAPH. For three reasons :

(1) you do not have anymore to consider vertical relationships. We once wondered if horizontal relations could emerge from a tree of only vertical ones, now we have the proof that the reverse is true : vertical relations emerge from a rhizome of only horizontal ones…

(2) you are not interested in individual nodes, but in PATHS, in NEIGHBORHOODS, and so on.

(3) Nodes themselves exist only with respect to a weight, as the “greatest coherence path” algorithm shows it… Thus, you don’t think of the node X, but of the node X times 1.56 for example.

And even that, but indeed, you see that the discrete segmentation of the structure into nodes is no longer needed. You could do the same algorithms with a CONTINUUM of sites.

Thus one yield to a new structure which I call a RHIZOME, in which the basic object is not the node X, but the “seme” X*1.56+Y*5.3+Z*2.01+…+K*3, a seme being a generalization of “node”, “path” and “vector”. I do not give you the exact mathematical definition, here I try to give you only the main idea…

So the course of history is from 100% LOGICISM to 100% HOLISM, through the following data structures :

LIST FILES —: TREES —: FIBRAGES —: DIRECTED GRAPHS —: NETS —: RHIZOMES —: ???

Hope this is an answer to your question – NicolasMontessuit

thanks for this good explanation.
situation seems as follows:
LIST FILES —: TREES —: FIBRAGES —: DIRECTED GRAPHS —: NETS —: RHIZOMES —:SENTENCES
sentences are results of rhizomes of words. you also can have a rhizome of different pictures and the result is an idea under the shape of a term. the brain of a speaking child is connecting these words to a new sentence (perhaps in the way, you described above) and so, you receive in a more logical sense a xanadu version:

 ill -> 1
 to heal -> 2, 6
 to mend -> 2, 3
 bugged -> 3, 4
 a robot -> 4
 somebody -> 5
 hospital -> 6
 ...
 *1 ill <does> to heal
 *2 to heal <is> to mend
 *3 bugged <is> to mend
 *4 a robot <does> bugged
 *5 somebody <is> ill
 *6 to heal <has> hospital
 ...
 from *4 & *3 you may conclude: 
 *7 a robot <does> *3
 from *7 & *2 you may conclude:
 *8 bugged <is> *2
 *9 a robot <does> *8
 from *9 & *12 you may conclude:
 *1o hospital <does> to heal
 *11 *10 <is> to mend
 *12 bugged <is> *11
 *16 a robot <does> *12
 that is: (a robot) <does> (bugged <is> (hospital <does> (to  heal)) <is> (to mend))
 or: (a robot) <is> (hospital <does> bugged) 
 . . ill
     d to  heal
   . to  heal
     i to  mend
   . bugged
     i to  mend
   . a  robot
     d bugged
   . some  body
     i ill
   . to  heal
     h hospital
 ~ a  robot
   d bugged
     i hospital
       d to  heal
     i to  mend

EditNearLinks: WikiWord BlueMoon NicolasMontessuit PatrickAnderson MotWiki

Languages: