Difference between revisions of "User:Unhammer"

From Apertium
Jump to navigation Jump to search
 
(15 intermediate revisions by the same user not shown)
Line 3: Line 3:
 
In Apertium, I work on
 
In Apertium, I work on
 
* [[Northern Sámi and Norwegian]], (which uses [[HFST]] for analysis),
 
* [[Northern Sámi and Norwegian]], (which uses [[HFST]] for analysis),
* [[Norwegian Nynorsk and Norwegian Bokmål]], with support from GsoC (see [[User:Unhammer/Application|project application]]),
+
* [[Norwegian Nynorsk and Norwegian Bokmål]], originally with support from GsoC (see [[User:Unhammer/Application|project application]]), lately as part of NTB/NPK's [https://www.nm.no/app/uploads/2020/03/nt-02-19.pdf#page=17 Nynorskroboten],
 
* [[Lt-trim]] and [[Emacs]] modes,
 
* [[Lt-trim]] and [[Emacs]] modes,
 
* [[Apy]]
 
* [[Apy]]
Line 9: Line 9:
 
* and have dabbled with [[Faroese and English]] (using [[Matxin]] and [[CG]]), Georgian, Dutch, Maltese, …
 
* and have dabbled with [[Faroese and English]] (using [[Matxin]] and [[CG]]), Georgian, Dutch, Maltese, …
   
I've studied computational linguistics / NLP at the University of Bergen, developed grammar checkers and Norwegian WordNets for Kaldera språkteknologi AS, and worked on Saami machine translation and corpus crawling for the University of Tromsø.
+
I've studied computational linguistics / NLP at the University of Bergen, developed grammar checkers and Norwegian WordNets for Kaldera språkteknologi AS, and worked on Saami grammar checking, machine translation and corpus crawling for the University of Tromsø.
   
 
Me on the web:
 
Me on the web:
 
* [[IRC]] nick: unhammer (most likely to reply between 8 and 15 CET)
 
* [[IRC]] nick: unhammer (most likely to reply between 8 and 15 CET)
* Google Melange link_id: unhammer
+
* Google/GsoC/GCI link_id: unhammer
  +
* [https://unhammer.org/k home page]
 
* [https://unhammer.wordpress.com dusty old blog]
 
* [https://unhammer.wordpress.com dusty old blog]
* [https://identi.ca/unhammer microblog]
+
* [https://github.com/unhammer GitHub]
* [https://www.openhub.net/accounts/unhammer OpenHUB] and [https://github.com/unhammer GitHub]
 
 
* [https://nn.wikipedia.org/wiki/Brukar:Unhammer nn.wikipedia page]
 
* [https://nn.wikipedia.org/wiki/Brukar:Unhammer nn.wikipedia page]
 
* [[Special:Emailuser/Unhammer|email]] me
 
* [[Special:Emailuser/Unhammer|email]] me
Line 47: Line 47:
 
<Unhammer> back to sme-nob, hopefully averting more water damage
 
<Unhammer> back to sme-nob, hopefully averting more water damage
 
</pre>
 
</pre>
  +
  +
<pre>
  +
<Claude_Royet-Journoud> Une liste d'infinitifs prolonge l'accident.
  +
</pre>
  +
  +
<center>But how powerful, how stimulating to the very faculty that produced it, <br/>
  +
was the invention of the adjective: no spell or incantation in Faerie is more potent.<br/>
  +
–J.R.R. Tolkien
  +
</center>
   
 
<pre>
 
<pre>
Line 53: Line 62:
 
<miri> I hope there is no correlation ;)
 
<miri> I hope there is no correlation ;)
 
</pre>
 
</pre>
  +
   
 
<pre>
 
<pre>
Line 58: Line 68:
 
<Unhammer> blasphemy
 
<Unhammer> blasphemy
 
</pre>
 
</pre>
  +
   
 
<center>
 
<center>
Line 64: Line 75:
 
– JJ
 
– JJ
 
</center>
 
</center>
  +
   
 
<center>
 
<center>
Line 72: Line 84:
   
 
IRC looks much better with some [http://www.vidarholen.net/contents/rage/ rage].
 
IRC looks much better with some [http://www.vidarholen.net/contents/rage/ rage].
  +
  +
<blockquote>
  +
Er man nihilistisk nok, kunne man også ta det uskyldigste av alle ord, infinitivsmerket, og misbruke og skjende det på denne måte: Det begynte «å» regne. Kan man se på verden med mindre begeistring?
  +
  +
–Bjørneboe
  +
</blockquote>
  +
  +
==Compounding is fun==
  +
  +
lemurtvillingene: lem|urt|villingene
  +
nyrestaurert: nyre|staur|ert
  +
angrepsoppstillinger: angrep|sopp|stillinger
  +
snusleverandør: snus|leve|rand|ør
  +
  +
$ echo bildreportagen |apertium -d . swe-dan
  +
billede #rids mugmide tagene
  +
  +
  +
^einannan/ein<n><m><sg><ind><cmp>+nan<n><m><sg><ind><cmp>+nan<n><m><sg><ind>$
  +
  +
$ echo nannannannannan|apertium -d . nno-nob-morph
  +
^nannannannannan/nan<n><m><sg><ind><cmp>+nan<n><m><sg><ind><cmp>+nan<n><m><sg><ind><cmp>+nan<n><m><sg><ind><cmp>+nan<n><m><sg><ind>$^./.<sent><clb>$
  +
  +
  +
$ echo fornybart|apertium -d . nob-dan
  +
fodernyoverskæg
  +
  +
Let's try compounding on verb+noun:
  +
  +
$ echo regionspresident | apertium -d . nob-nno_e
  +
region spreie seie dent
  +
  +
noun+verb:
  +
  +
$ echo forringelse | apertium -d . nob-nno_e
  +
for ringel sjå
  +
  +
noun+verb+adj:
  +
  +
$ echo autoritært|lt-proc -we nob-dan.automorf.bin
  +
^autoritært/auto<n><mf><sp><cmp>+ri<vblex><inf><cmp>+tære<adj><pp><pl>/auto<n><mf><sp><cmp>+ri<vblex><inf><cmp>+tære<adj><pp><nt><sg><ind>/auto<n><mf><sp><cmp>+ri<vblex><inf><cmp>+tære<adj><pp><mf><sg><ind>$
  +
  +
  +
What if we allow turning double consonants into single before the compound border, then we can analyse compounds of words ending in double consonants followed by a word starting with the same consonant:
  +
  +
$ echo topprøve|apertium -d . nob-nno_e-morph|cg-conv
  +
"<topprøve>"
  +
"røve" adj pp pl
  +
"topp" n m sg ind cmp
  +
"prøve" n m sg ind
  +
"topp" n m sg ind cmp detriple
  +
  +
but of course the analyser decompounding doesn't know that the second word has to actually start with that same consonant:
  +
  +
$ echo HurtigrutenLive|apertium -d . nob-nno_e
  +
hurr TigruteinLive
   
 
[[Category:Users]]
 
[[Category:Users]]

Latest revision as of 18:11, 12 December 2022

I am Kevin Brubeck Unhammer.

In Apertium, I work on

I've studied computational linguistics / NLP at the University of Bergen, developed grammar checkers and Norwegian WordNets for Kaldera språkteknologi AS, and worked on Saami grammar checking, machine translation and corpus crawling for the University of Tromsø.

Me on the web:

I have an Apertium /wishlist.

♪ Unhaaamer Unhaaamer, He beat the Hun by luck.

Unhaaamer Unhaaamer, he's smarter than a duck ♪

Quotes[edit]

They've a temper, some of them—particularly verbs: they're the proudest
—adjectives you can do anything with, but not verbs—however, I can
manage the whole lot of them! Impenetrability! That's what I say!
— Humpty Dumpty

<Unhammer> Every time I start working on a new Apertium lang. pair, I get water damage in my apartment.
<spectie> Unhammer, are you sure you want to start working on ht-en
<spectie> what with your precarious plumbing situation ?
...
<Unhammer> back to sme-nob, hopefully averting more water damage
<Claude_Royet-Journoud> Une liste d'infinitifs prolonge l'accident.
But how powerful, how stimulating to the very faculty that produced it,

was the invention of the adjective: no spell or incantation in Faerie is more potent.
–J.R.R. Tolkien

  <miri> now the internet is back
  <miri> but there's no water in my building
  <miri> I hope there is no correlation ;)


<Unhammer>  [-#Ipmil-] {+Ipmil+}
<Unhammer> blasphemy


the warm soft short pants of the quick-scribbler: the vocative lapse from which it begins and the accusative hole in which it ends itself

– JJ


There are a number of languages spoken by human beings in this world.

– Harald Tveit Alvestrand, in RFC 1766, "Tags for the Identification of Languages"

IRC looks much better with some rage.

Er man nihilistisk nok, kunne man også ta det uskyldigste av alle ord, infinitivsmerket, og misbruke og skjende det på denne måte: Det begynte «å» regne. Kan man se på verden med mindre begeistring?

–Bjørneboe

Compounding is fun[edit]

   lemurtvillingene: lem|urt|villingene
   nyrestaurert: nyre|staur|ert
   angrepsoppstillinger: angrep|sopp|stillinger
   snusleverandør: snus|leve|rand|ør
   $ echo bildreportagen |apertium -d . swe-dan
   billede #rids mugmide tagene


   ^einannan/ein<n><m><sg><ind><cmp>+nan<n><m><sg><ind><cmp>+nan<n><m><sg><ind>$
   …
   $ echo nannannannannan|apertium -d . nno-nob-morph
   ^nannannannannan/nan<n><m><sg><ind><cmp>+nan<n><m><sg><ind><cmp>+nan<n><m><sg><ind><cmp>+nan<n><m><sg><ind><cmp>+nan<n><m><sg><ind>$^./.<sent><clb>$


   $ echo fornybart|apertium -d . nob-dan
   fodernyoverskæg

Let's try compounding on verb+noun:

   $ echo regionspresident | apertium -d . nob-nno_e
   region spreie seie dent

noun+verb:

   $ echo forringelse | apertium -d . nob-nno_e
   for ringel sjå

noun+verb+adj:

   $ echo autoritært|lt-proc -we nob-dan.automorf.bin
   ^autoritært/auto<n><mf><sp><cmp>+ri<vblex><inf><cmp>+tære<adj><pp><pl>/auto<n><mf><sp><cmp>+ri<vblex><inf><cmp>+tære<adj><pp><nt><sg><ind>/auto<n><mf><sp><cmp>+ri<vblex><inf><cmp>+tære<adj><pp><mf><sg><ind>$


What if we allow turning double consonants into single before the compound border, then we can analyse compounds of words ending in double consonants followed by a word starting with the same consonant:

   $ echo topprøve|apertium -d . nob-nno_e-morph|cg-conv
   "<topprøve>"
       "røve" adj pp pl
               "topp" n m sg ind cmp
       "prøve" n m sg ind
               "topp" n m sg ind cmp detriple

but of course the analyser decompounding doesn't know that the second word has to actually start with that same consonant:

   $ echo HurtigrutenLive|apertium -d . nob-nno_e
   hurr TigruteinLive