Improved Sanskrit Computational Linguistics Platform

[Gérard Huet]

Dear colleagues,

this message is to inform you of the recent release of a significantly enhanced version of the Sanskrit Heritage platform at http://sankrit.inria.fr

Main improvements:
1. The Sanskrit Reader is now driven through a graphical interface giving a concise summary of segmentation solutions. This allows the input
of more realistic Sanskrit sentences, without the risk of choking the server with enormous Web pages. Furthermore, it is now much more convenient to
focus rapidly on the solution by clicking on preferred segments. Each segment may be inspected, revealing its morphological parameters and lemma,
with possible lexicon access. At any point the user may undo his moves, or switch to the old interface when few potential solutions remain.
2. When an input chunk is not analysable by the system, the user is offered the possibility to modify it or suppress it.
3. Input is possible in 4 ASCII transliteration schemes, as well as in Unicode UFT-8, either for romanized transcription with diacritics (IAST) or for devanāgarī.
4. Lexicon access is possible to the Sanskrit to French Heritage dictionary, as before, but also to an HTML version of the Monier-Williams English dictionary. Sticky preference to this last choice is encapsulated in the English entry page http://sanskrit.inria.fr/index.en.html
5. Numerous fixes to the morphology generation schemes have been effected, with a view to conform to Paninian processes.

All comments welcome
G. Huet & P. Goyal