On 3 October 2017 at 22:21, Lorenza Russo <lorenzaru...@hotmail.com> wrote: > Hello, > > I am writing to bring to your attention some inconsistencies I have found in > some pages about cross dix. > > It's the first time that I do cross dics, so I followed the examples I found > on the wiki platform, and thanks also to the error messages I got from the > program, I could spot two kind of inconsistencies: > > 1) typos/wrong information in the examples given; > 2) typos/wrong information in the explanations given. >
Well, yes and no. There are inaccuracies, but for the most part, they are not the things you mention. > If something it's not clear, please don't hesitate to let me know. If I am > wrong about some of my remarks, sorry for this spam. As I said, I am > learning by doing. > > Let's start with the files given as examples: > > a) The "Example of LRD" at this page: > http://wiki.apertium.org/wiki/Linguistic_Resources_Document (all the way > down the page) seems to be an example for creating English-Spanish > dictionaries by crossing English-Catalan and Catalan-Spanish. I couldn’t > find (in trunk, at least) any Catalan-Spanish but I don't know if the couple > of language being in trunk is something needed for the example to be > efficient. For the rest: The pair is Spanish-Catalan; in this case, dixtools reverses the Spanish-Catalan bilingual dictionary. Using an LRD is supposed to let dixtools figure out what to do itself; I'm not sure if the cross-dics tool actually works with an LRD, though. > > 1) at line 19 and at line 30, the cross model should be > "cross-model-en-ca-es.xml" instead of "cross-model-es-ca-en.xml" It doesn't really matter what this is called, it's just a name. > 2) I am not 100% sure here, but maybe the order of the morphological > dictionaries should be inverted, first the morphological dictionary for > English and then the one for Spanish The order of the morphological dictionaries shouldn't matter. > 3) the bilingual dictionary starting from line 67 should be rewritten > inverting the order of the languages. Now it is like this: > <property name="name" value="apertium-es-ca"/> > <property name="type" value="bil"/> > <property name="sl" value="es"/> > <property name="tl" value="ca"/> > <property name="for-crossing" value="yes"/> > <property name="src" value="apertium-es-ca.es-ca.dix"/> > <property name="version" value="stable"/> > but it should be > <property name="name" value="apertium-ca-es"/> > <property name="type" value="bil"/> > <property name="sl" value="ca"/> > <property name="tl" value="es"/> > <property name="for-crossing" value="yes"/> > <property name="src" value="apertium-ca-es.ca-es.dix"/> > <property name="version" value="stable"/> > No: the bilingual dictionary in the apertium-es-ca package is apertium-es-ca.es-ca.dix, not apertium-ca-es.ca-es.dix. The provided values are correct here. > > b) Let's consider now the example available here: > http://wiki.apertium.org/wiki/Crossdics_Example for creating > English-Galician by crossing English-Spanish and Spanish-Galician (see the > example of en-es-gl-ling-resources.xml almost at the end of the page): > > 1) the morphological dictionary at line 38, should be > "apertium-en-es.es.metadix" and not "apertium-en-es.en.metadix" > You mean, in this section? <!-- 'en' morphological dictionary --> <resource> <property name="name" value="apertium-en"/> <property name="type" value="mon"/> <property name="sl" value="en"/> <property name="for-crossing" value="yes"/> <property name="src" value="apertium-en-es.en.metadix"/> </resource> This is describing the 'en' morphological dictionary, it absolutely should not be "apertium-en-es.es.metadix" (which, as it happens, does not exist). The idea seems to have been to trim the source and target monolingual dictionaries to remove monolingual entries that did not have corresponding bilingual entries in the crossed dictionary; this has never worked in my experience, but even if it did, there would be no use for the Spanish monolingual dictionary, as it is the pivot language in this example. > As for the inconsistencies about the explanations given, I found the > following ones: > > > 1) at the page http://wiki.apertium.org/wiki/Crossdics_Example in section > "What you need" and then "Dictionaries" there is a typo for the > Morphological dix A: now it's > > Morphological A: apertium-en-es.en.dix > but I think it should be > Morphological A: apertium-en-es.es.dix There is an error, but it's that the filename ought to be apertium-en-es.en.metadix - there is no file named "apertium-en-es.en.dix". I've corrected this. > 2) at the page http://wiki.apertium.org/wiki/Crossdics in section "Without a > linguistic resources document", the scheme for the dix is inverted. Now it's > > Bilingual dictionary A-B: apertium-bb-aa.bb-aa.dix > Morphological dictionary A: apertium-bb-aa.aa.dix > > but to be aligned to what it's written in other pages it should be > > Bilingual dictionary A-B: apertium-aa-bb.aa-bb.dix > Morphological dictionary A: apertium-aa-bb.bb.dix Ok, the model, as given, was incorrect: Bilingual dictionary A-B: apertium-bb-aa.bb-aa.dix Bilingual dictionary B-C: apertium-bb-cc.bb-cc.dix Morphological dictionary A: apertium-bb-aa.aa.dix Morphological dictionary C: apertium-bb-cc.cc.dix I've corrected it to say B-A, not A-B. The reason is given a little lower on the page: 'the common language (B) must be in the left side, that is, dictionaries in the form B-A and B-C' If it happens that your bilingual dictionaries don't follow this model: if you have A-B or C-B or both, you can follow the instruction on the next line: 'use "-r" instead of "-n" if the dictionary has to be reversed (apertium-aa-bb.aa-bb.dix to apertium-bb-aa.bb-aa.dix)' ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ Apertium-stuff mailing list Apertium-stuff@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/apertium-stuff
