Difference between revisions of "Lttoolbox-java/Flag diacritics"

From Apertium
Jump to navigation Jump to search
(Category:Documentation in English)
Line 93: Line 93:
   
 
http://apertium.svn.sourceforge.net/viewvc/apertium/trunk/lttoolbox-java/testdata/flag_matching/persian.dix?view=markup
 
http://apertium.svn.sourceforge.net/viewvc/apertium/trunk/lttoolbox-java/testdata/flag_matching/persian.dix?view=markup
 
   
 
http://apertium.svn.sourceforge.net/viewvc/apertium/trunk/lttoolbox-java/testdata/flag_matching/persian2.dix?view=markup
 
http://apertium.svn.sourceforge.net/viewvc/apertium/trunk/lttoolbox-java/testdata/flag_matching/persian2.dix?view=markup

Revision as of 15:16, 31 May 2012

This describes an experimental mode for lttoolbox-java which was made in december 2010 but never brought into use.

The idea is to add 'flag' symbols that must match in the lttoolbox expansion.

Given the symbols

<sdef n="mi:1" /> 
<sdef n="mi:0" />
<sdef n="bi:0" />
<sdef n="bi:1" />

A given expansion will be pruned away if there are conflicting flag symbols. For example an expansion with both <mi:1> and <mi:0> will be pruned awway.

A <mi:1> and <bi:0> won't be pruned away as these flags doesen't conflict.

Here is an example:

<pardef n="prefix">
  <!-- 'bi-' is the prefix for indicative -->
  <e>       <p><l>bi</l>        <r><s n="bi:1"/></r></p></e>

  <!-- 'mi-' is the prefix for subjunctive -->
  <e>       <p><l>mi</l>        <r><s n="mi:1"/></r></p></e>

  <e>       <p><l></l>          <r><s n="xp:1"/></r></p></e>
</pardef>

<pardef n="khoda/n__vblex">
  <e>       <p><l>m</l>         <r>n<s n="xp:0"/><s n="mi:1"/><s n="bi:0"/><s n="vblex"/><s n="pri"/><s n="p1"/><s n="sg"/></r></p></e>
  <e>       <p><l>m</l>         <r>n<s n="xp:0"/><s n="mi:0"/><s n="bi:1"/><s n="vblex"/><s n="prs"/><s n="p1"/><s n="sg"/></r></p></e>

  <e>       <p><l>š</l>         <r>n<s n="xp:0"/><s n="mi:1"/><s n="bi:0"/><s n="vblex"/><s n="pri"/><s n="p2"/><s n="sg"/></r></p></e>
  <e>       <p><l>š</l>         <r>n<s n="xp:0"/><s n="mi:0"/><s n="bi:1"/><s n="vblex"/><s n="prs"/><s n="p2"/><s n="sg"/></r></p></e>

  <e>       <p><l></l>          <r>n<s n="xp:0"/><s n="mi:1"/><s n="bi:0"/><s n="vblex"/><s n="pri"/><s n="p3"/><s n="sg"/></r></p></e>
  <e>       <p><l></l>          <r>n<s n="xp:0"/><s n="mi:0"/><s n="bi:1"/><s n="vblex"/><s n="prs"/><s n="p3"/><s n="sg"/></r></p></e>
  <e>       <p><l>n</l>         <r>n<s n="xp:1"/><s n="vblex"/><s n="inf"/></r></p></e>
</pardef>


Now, if the 2 pardefs are combined in, say


<e lm="khodan">          <par n="prefix"/><i>khoda</i><par n="khoda/n__vblex"/></e>

we don't get all the 3x7 possibilites because conflicting flags are ruled out.

Effectively we get only:

bikhodam:<bi:1>khodan<mi:0><xp:0><vblex><pri><p1><sg>
bikhodaš:<bi:1>khodan<mi:0><xp:0><vblex><pri><p2><sg>
bikhoda:<bi:1>khodan<mi:0><xp:0><vblex><pri><p3><sg>
mikhodam:<mi:1>khodan<bi:0><xp:0><vblex><prs><p1><sg>
mikhodaš:<mi:1>khodan<bi:0><xp:0><vblex><prs><p2><sg>
mikhoda:<mi:1>khodan<bi:0><xp:0><vblex><prs><p3><sg>
khodan:<xp:1>khodan<xp:1><bi:0><mi:0><vblex><inf>

The lttoolbox-java processor automatically removes flag symbols from the output, so in the end we get just

bikhodam:khodan<vblex><prs><p1><sg>
bikhodaš:khodan<vblex><prs><p2><sg>
bikhoda:khodan<vblex><prs><p3><sg>
mikhodam:khodan<vblex><pri><p1><sg>
mikhodaš:khodan<vblex><pri><p2><sg>
mikhoda:khodan<vblex><pri><p3><sg>
khodan:khodan<vblex><inf>


Usage

Here are the relevant flags for lttoolbox-java processor.

$ lt-proc-j 
  -f:   match flags (experimental)
  -S:   show hidden control symbols (for flagmatch and compounding)


Further reading

http://apertium.svn.sourceforge.net/viewvc/apertium/trunk/lttoolbox-java/test/org/apertium/lttoolbox/FlagMatchingTest.java?view=markup

http://apertium.svn.sourceforge.net/viewvc/apertium/trunk/lttoolbox-java/testdata/flag_matching/persian.dix?view=markup

http://apertium.svn.sourceforge.net/viewvc/apertium/trunk/lttoolbox-java/testdata/flag_matching/persian2.dix?view=markup