ponteineptique/deucalion-model-af repository overview

⁠Deucalion Model Ancien Français ENC

⁠Credits

This software is based on Pie by Enrique Manjavacas (@emanjavacas) and Mike Kestemont (@mikekestemont)

The web application and its maintenance is done by Thibault Clérice ( @ponteineptique ). To learn how to cite this repository, go check our releases⁠.

⁠Information about the model

The model was trained on the following corpora :

• Geste: un corpus de chansons de geste, dir. Jean-Baptiste Camps, avec la collab. d'Elena Albarran, Alice Cochet & Lucence Ing, Paris, 2016-…, http://github.com/Jean-Baptiste-Camps/Geste⁠.
• Édition nativement numérique du recueil hagiographique "Li Seint Confessor" de Wauchier de Denain d'après le manuscrit 412 de la Bibliothèque nationale de France, éd. Ariane Pinche, Lyon, en cours (only Dialogues and La vie de Saint Martin are used right now. Data are closed source until publication of the PhD thesis)
• Les Institutes de Justinien en français, éd. F. Olivier-Martin (1935), éd. revue par F. Duval, lemmatisée par F. Duval et L. Ing. Paris, 2018.
• Chrétien de Troyes: Cligès, Erec, Lancelot, Perceval, Yvain -- Manuscrit P (BnF Fr. 794), éd. P. Kunstmann (2009), annotation revue par J.B. Camps et L. Ing (2017).

The annotations are made according to the following reference lists:

• lemma: Adolf Tobler et Erhard Friedrich Lommatzsch, Altfranzösisches Wörterbuch: édition électronique, éd. Peter Blumenthal et Achim Stein, Stuttgart, F. Steiner, 2002.
• POS and morph: Sophie Prévost, Céline Guillot, Alexei Lavrentiev et Serge Heiden, Jeu d’étiquettes morphosyntaxiques CATTEX2009-max, Lyon, 2013, http://bfm.ens-lyon.fr/IMG/pdf/Cattex2009_2.0.pdf⁠.

More information on the annotation practice can be found in the wiki of the Geste corpus: https://github.com/Jean-Baptiste-Camps/Geste/wiki⁠.

Sample from annotation:

form	lemma	POS	morph
G'	je	PROper	PERS.=1|NOMB.=s|GENRE=m|CAS=n
irai	aler	VERcjg	MODE=ind|TEMPS=fut|PERS.=1|NOMB.=s
sor	sor2	PRE	MORPH=empty
eus	il	PROper	PERS.=3|NOMB.=p|GENRE=m|CAS=i
por	por2	PRE	MORPH=empty
lor	lor2	DETpos	PERS.=3|NOMB.=p|GENRE=f|CAS=r
terres	terre	NOMcom	NOMB.=p|GENRE=f|CAS=r
saisir	saisir	VERinf	MORPH=empty

⁠Error you can expect

If you want to have an idea about the errors the model can do, please have a look at this file for lemma and POS⁠ and this file for morph⁠

⁠Example

This was run on a text that was not seen during training :

⁠Install

To run, you'll need to install Docker. Then, you can simply run the following commands :

docker build -t deucalion-af:latest .
docker run -p 5001:5000 deucalion-af:latest

You can replace 5001 with any port you want to run the service on.

Then, simply go to http://127.0.0.1:5001⁠ or directly to http://127.0.0.1:5001/api?data=SOIGNORS,or%20escoutez,que%20D%C3%A9s%20vos%20soit%20amis,.III%20.vers%20de%20bone%20estoire,se%20je%20les%20vos%20devis⁠ or post any data to the same URI with data parameters containing your plain text.