Hi there, I have a proposal for GSOC but I do not know if publish it explicitly in the wiki page:
http://wiki.apertium.org/wiki/Ideas_for_Google_Summer_of_Code Premise: yes, I am still alive... Hu, well, after some tricky "political" things at my University (L'Aquila, Italy) me and Antonio Barbieri, my student in A.I. who is preparing a thesis in machine translation for his MA, finally came to a definition which can be a proposal for GSOC. After a skype conversation with Hector, I understood that one of the problem with Romance languages is the fact that many of them are pro-drop, and pronouns are even ambigous ("se" in Catalan, Spanish, Portoguese, which becomes "si" in Italian). Furthermore, words like for instance "comunista" can be both adjective ("un giornale comunista") or a noun ("un comunista arrabbiato") and this poses a lot of problems -- well-known in the MT literature, BTW. My idea is to implement my morphosyntactic tree model for disambiguation using Constraint Grammar as a start point, and at the same time starting the it-eo language pair, that I would maintain even after the ending of Antonio's MA thesis. For people interested, my model is called Constructive Adpositional Grammars (CxAdGrams) and you can download legally a sample of the book here: http://federicogobbo.name/en/cxadgrams.html However, I already found the camera-ready in non-legal sites, so if you want the PDF of the book feel free to ask. This is not an advertisement, in fact I gain no money in the book... :/ I know that Jacob and Hector can help me for the launching, and also Antonio Toral gave me good advices. So, what do you think? Should I publish this idea? Thanks a lot, sorry if I was so long in this email... Federico PS For my academic career, it would be nice for me to be a mentor, but I understand that I am a newbie so it is not a request: the important thing is that Antonio should finish his thesis by 1 October 2013. Perhaps one of the actual mentor can be interested in it. Of course, the review of the quality of translations it-eo is completely mine. Kara Federico, Mi ĵus retbabilis kun Francis Tyers (spectie) en IRC #apertium kaj rakontis pri la projekto. Li estas tre favora. Li havas siajn ideojn kaj ŝatus retbabili kun vi. Unu el la gravaj problemoj, kiujn ni havas por ĉiuj latinidaj lingvoj, estas la multsignifeco/multfunkcieco de ŝajnpronomaj verbaj formoj: la pronomo "se" de la kat-hisp-port-ktp. (tre certe ankaŭ itale) en esperanto povas fariĝi "si", "oni", nenio (ĉar ĝi nur plifortigas la signifon de la verbo), -iĝ k.a. Do, mi kredas, ke li ŝatus, ke vi pensu ankaŭ pri tio en la projekto. Krom tio li ja simple volas konatiĝi kun novuloj. (Cetere, li estas la ulo, kiu finskribas sian disertacion pri leksika elekto, kaj ankaŭ tiu, kiu verkis la plimulton de la Apertium-vikio, krom tio, ke li laboris super la plej diversaj lingvoj, ĉefe inter la malplej uzataj: bretona, islanda, kimra, sarda, nord-samea, tatara, ĉuvaŝa k.a.) Kore, Hektor -- Federico Gobbo, PhD Research Fellow in Computer Science Autonomous Agents & A.I. @AQ Research Group DISIM - Dip. di Ing. e Sc. dell'Informazione e Mat. via Vetoio - Coppito Zero - Room 209 IT-67100 L'Aquila (Italia) tel ++39 0862-433172 web http://federicogobbo.name ------------------------------------------------------------------------------ Everyone hates slow websites. So do we. Make your web apps faster with AppDynamics Download AppDynamics Lite for free today: http://p.sf.net/sfu/appdyn_d2d_mar _______________________________________________ Apertium-stuff mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/apertium-stuff
