apertium-cat-ita is ready for a release. Since my last message, I have added c. 1000 entries in the bidix, and improved a bit transfer and lexical selection rules. Testvoc has been passed again.
Hèctor ---------- Forwarded message --------- De: Hèctor Alòs i Font <hectora...@gmail.com> Date: dt., 23 de jul. 2019 a les 12:05 Subject: GSoC 2019: Catalan-Italian ready To: [apertium-stuff] <apertium-stuff@lists.sourceforge.net> My GSoC has two parts. In the first one, my task was to improve the translator from Italian to Catalan and build a new one in the opposite side. Both are ready. There are no testvoc errors. WER in the Italian to Catalan side is found to be 15,7% while in the Catalan-Italian side it is 14.2%. The better score for the second side (although newer) is the result of a more mature monolingual package, with a bigger dictionary and a more reliable disambiguation. I'll be glad if the package can be packaged and released. In short, these are the main improvements: * The Italian dictionary has been increased (in part, thanks to the help of Jaume Ortolà, as for other language pairs) * The Italian disambiguation has been largely improved * The Italian post-generation has been improved * The bilingual dictionary has been hugely increased * There are hundreds of the lexical selection rules * Transfer rules have been increased both at t1x and t2x levels Many disambiguation rules for Catalan have been reused from the Catalan package. Many lexical selection rules have been reused from the Catalan-Sardinian pair and, to a lesser extent, from the Catalan-French one. Transfer rules both at t1x and t2x levels reuse code from the French-Occitan pair. The main issues are: * More anthroponyms and, above all, toponyms would be fine * Still a few issues in the Italian post-edition are found * Still issues in disambiguation are found, especially for Italian * There are no transfer rules for switching from pronominal verbs to non-pronominal and the opposite * Lexical selection needs to better distinguish between transitive and intransitive verbs * Transfer rules need to better distinguish between transitive and intransitive verbs (mostly for electing the right auxiliary verb in Italian) * The Italian partitive article continues to give some problems (although in real texts less than I feared); transfer rules must be improved on both sides * The agreement between subject and attribute should be improved I thank my mentors, Xavi Ivars and Marc Riera, for the help they have given me, particularly Xavi with scripts, and Marc for his comments. Thanks also to Tino and Kevin for their help in several issues. And, as already said, to Jaume Ortolà for his extremely helpful language data from LanguageTool. I am especially indebted to Gianfranco Fronteddu, whom I have riddled with questions on Italian. He has also reviewed several Italian automatic translations giving me many valuable comments. Hèctor
_______________________________________________ Apertium-stuff mailing list Apertium-stuff@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/apertium-stuff