Jean-Baptiste-Camps / stemmatology

Stemmatological Analysis of Textual Traditions
GNU General Public License v3.0
15 stars 3 forks source link

PCC.reconstructModel : comparisons with witnesses outside the group #30

Open Jean-Baptiste-Camps opened 6 years ago

Jean-Baptiste-Camps commented 6 years ago

When comparing reconstructed models to extant witnesses, do we need to look outside the cluster ? It is included in the paper and maybe in Poole too, but it does not seem algorithmically consistent (except in cases without severe disagreements), because clusters will be made based on severe disagr. with other witnesses, and, as such, no outsider could be their model.

Comments in code:

        # NB & TODO(GLOBAL): this step (that we included in
        # the paper) is PROBABLY not necessary, nor algorithmically consistent.
        # How can the model be outside the group and
        # have no disagreement with the model, knowing that the virtual model is
        # reconstructed based on common readings to the mss of the group, and
        # that these are, at least once, unique to this group? Yet the complexity
        # of this principle is very high, and intuition hard, so we need to
        # check it.
floriancafiero commented 6 years ago

Pas sûr que ce soit un problème en fait, tu as un exemple pour illustrer l’erreur potentielle ?

Le 1 mai 2018 à 16:34, Jean-Baptiste-Camps notifications@github.com a écrit :

When comparing reconstructed models to extant witnesses, do we need to look outside the cluster ? It is included in the paper and maybe in Poole too, but it does not seem algorithmically consistent (except in cases without severe disagreements), because clusters will be made based on severe disagr. with other witnesses, and, as such, no outsider could be their model.

Comments in code:

    # NB & TODO(GLOBAL): this step (that we included in
    # the paper) is PROBABLY not necessary, nor algorithmically consistent.
    # How can the model be outside the group and
    # have no disagreement with the model, knowing that the virtual model is
    # reconstructed based on common readings to the mss of the group, and
    # that these are, at least once, unique to this group? Yet the complexity
    # of this principle is very high, and intuition hard, so we need to
    # check it.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

Jean-Baptiste-Camps commented 6 years ago

Pas forcément un problème en effet, c'est juste une étape coûteuse (temps de calcul, etc.), redondante dans le code, et pas forcément indispensable. Peux-tu imaginer une configuration dans lequel le modèle d'un groupe lui serait extérieur ? Ça me paraît impossible, mais il faudrait le prouver.

floriancafiero commented 6 years ago

Oui c’est juste, c’est coûteux algorithmiquement, mais pas forcément dommageable en dehors de ca.

Pas réussi à trouver d’exemple non plus en effet, c’est pour ça que je te demandais :) Mais ça ne prouve rien, dès que j’ai fini mes clowneries (et CornMol), je regarde si on peut démontrer ça proprement.

Le 2 mai 2018 à 10:00, Jean-Baptiste-Camps notifications@github.com a écrit :

Pas forcément un problème en effet, c'est juste une étape coûteuse (temps de calcul, etc.), redondante dans le code, et pas forcément indispensable. Peux-tu imaginer une configuration dans lequel le modèle d'un groupe lui serait extérieur ? Ça me paraît impossible, mais il faudrait le prouver.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.

Jean-Baptiste-Camps commented 6 years ago

Je pense que oui:

  1. les groupes sont constitués en fonction des désaccords graves qui les opposent à d'autres manuscrits;
  2. cela signifie donc que, dans un même groupe, les manuscrits s'accordent sur des leçons qui les opposent au reste de la tradition;
  3. s'ils possèdent des leçons de ce type, c'est qu'ils les ont héritées d'un ancêtre commun.

Ergo: le modèle, s'il est conservé, est forcément dans le groupe.

Après, ce qui me fait hésiter, c'est la complexité: comme les groupes se forment à des étapes différentes, et que des choses peuvent s'être résolues entretemps, etc., on pourrait imaginer que ça influe sur ce principe…

Jean-Baptiste-Camps commented 6 years ago

Un contre-exemple à explorer, stemma_with_model_outside-group