Difference between revisions of "Polished RBMT system"

From LING073
Jump to: navigation, search
m (The assignment)
m (The assignment)
 
(28 intermediate revisions by 2 users not shown)
Line 1: Line 1:
== Hand-annotating corpora ==
+
''If you're looking for information on corpus annotation and evaluation (precision/recall), that has moved to the [[Midterm overview]] page''
First you want to analyze your corpus and output to CG format:
 
cat corpus.txt | apertium -d . xyz-morph | cg-conv -a > corpus.out.txt
 
  
Your new file probably now looks something like this:
+
== Measuring trimmed coverage ==
<pre>
 
"<This>"
 
"this" det dem sg
 
"this" prn dem sg
 
"<is>"
 
"be" vbser pres p3 sg
 
"<my>"
 
"I" prn p1 sg pos
 
"<house>"
 
"house" n sg
 
"<.>"
 
"." sent
 
"<I>"
 
"I" prn p1 sg subj
 
"<live>"
 
"live" vblex inf
 
"live" vblex past
 
"<here>"
 
"*here"
 
"<..>"
 
".." sent
 
</pre>
 
  
In this example, you might note a few fixes:
+
Apertium RBMT systems use a "trimmed" version of the base transducer for each language in the MT pair.  Each transducer is trimmed to only include entries found in the bilingual dictionary (<code>.dix</code>).  The trimmed transducers are almost always smaller than the main transducers.  Measuring coverage of the trimmed transducer gives you some idea of the ceiling of MT accuracy—or, at least, the percentage of forms in a corpus the MT system can analyse, regardless of translation accuracy.
* "here" isn't being analysed; it should have an adverb reading
 
* "house" should have a verb reading
 
* "live" should have an adjective reading
 
* "live" isn't the past tense form of this verb
 
  
The following annotation makes these corrections:
+
Measuring trimmed coverage is just the same as measuring coverage, but with the appropriate "trimmed" transducer (e.g., <code>xyz-abc.automorf.bin</code>).  You'll need to use the <code>coverage-ltproc</code> script instead of <code>coverage-hfst</code>.
 
 
<pre>
 
"<This>"
 
"this" det dem sg
 
"this" prn dem sg
 
"<is>"
 
"be" vbser pres p3 sg
 
"<my>"
 
"I" prn p1 sg pos
 
"<house>"
 
"house" n sg
 
"house" vblex tv inf
 
"<.>"
 
"." sent
 
"<I>"
 
"I" prn p1 sg subj
 
"<live>"
 
"live" vblex inf
 
"live" adj
 
"<here>"
 
"here" adv
 
"<..>"
 
".." sent
 
</pre>
 
 
 
Note: There should be no unknown words ("analyses" with *) when you're done.
 
 
 
== Measuring precision and recall ==
 
[[:wikipedia:Precision and recall|Precision and recall]] are measures of how accurate a transducer is.  Precision is the number of returned analyses that are correct, and recall is the number of correct analyses that are returned.
 
 
 
In the example above of hand annotation, the precision is 90% (there are 9 true positives and 1 false positive), meaning that 90% of the returned analyses were correct. Recall is lower, at 75% (there are 9 true positives and 3 false negatives), meaning that only 75% of the correct analyses were returned.
 
 
 
There is a script in the [[Using the tools on your own system#Misc tools|tools repo]] called <code>precisionRecall</code>.  You can update the repo (<code>git pull</code>) and run <code>sudo make</code> to ensure that you have this script installed on your system.  You can then run <code>precisionRecall referencecorpus.txt annotatedcorpus.txt</code>.
 
  
 
== The assignment ==
 
== The assignment ==
 +
This assignment is due at the end of week 12 (this semester, at the end of the day on '''Friday, 29 April 2022, before midnight''').
  
 
# '''Before you begin''', make sure all previous assignments are done, and add a "structural_transfer" tag to your transducer repositories and your translation pair repository/ies to mark the end of previous assignments.
 
# '''Before you begin''', make sure all previous assignments are done, and add a "structural_transfer" tag to your transducer repositories and your translation pair repository/ies to mark the end of previous assignments.
 +
#* Also, please remove all binaries from all repositories!  See [[removing binaries from transducer repo]].
 
# Set up some '''new corpora''' based on existing ones:
 
# Set up some '''new corpora''' based on existing ones:
#* Combine your <code>sentences</code> and <code>tests</code> corpora so you have a new '''longer parallel corpus'''.  Name the files <code>abc.longer.txt</code> and <code>xyz.longer.txt</code>.
+
#* Combine and merge your <code>sentences</code> and <code>tests</code> corpora so you [hopefully, but not necessarily] have a new '''longer parallel corpus'''.  Name the files <code>abc.longer.txt</code> and <code>xyz.longer.txt</code>.
#* Make a '''large monolingual corpus''' of a bunch of raw text in your language.  The more the better.  This step may simply consist of you cleaning up and/or combining the existing corpora from the [[initial corpus assembly]] assignment.  See if you can get it over 100K words.  The bigger this corpus is the better.  Call it <code>abc.corpus.large.txt</code> (in your monolingual corpus repo) and add notes to your <code>MAINFEST</code> file about where the text comes from.
+
#* Make a '''large monolingual corpus''' of a bunch of raw text in your language.  The more the better.  This step may simply consist of you cleaning up and/or combining the existing corpora from the [[initial corpus assembly]] assignment.  See if you can get it over 100K words.  The bigger this corpus is the better.  Call it <code>abc.corpus.large.txt</code> (in your monolingual corpus repo) and add notes to your <code>MANIFEST</code> file about where the text comes from.
#* A '''hand-annotated monolingual corpus''' of sentences (see above) covering at least 1000 characters (500 for syllabic scripts) of your <code>abc.corpus.basic.txt</code> file, ideally sentences you understand / have English glosses of.  Put the sentences you want to annotate in <code>abc.annotated.raw.txt</code> and dump this to <code>abc.annotated.basic.txt</code> to annotate it in CG format.  Add these files to your monolingual corpus repository.
+
# '''Expand your MT pair''' in at least '''three''' of the following ways for each translation direction that you're working on, listing in a "Final evaluation" section on the language pair's wiki page what you did (move existing evaluation sections under a new section called "Initial evaluation"), and for every rule (for all of the following except adding stems), list an example of what output was improved.
# If you've been working on separate MT pairs, '''combine your MT pairs''' into one repository (which you both have full access to), making sure to incorporate all of the following:
+
#* At least '''100 more stems''' in the bilingual dictionary (and monolingual dictionaries as needed). This counts for both translation directions, if you are working on two-way translation.  (If you've automated creation of your bilingual dictionary, then you can use this task to clean up 100 stems—at least 50 entries in your bilingual dictionary should be different than before—and properly categorise those 100 stems into the monolingual dictionary.)
#* All entries from both dictionaries in a single <code>.dix</code> file.  Make sure all translations are in the default direction of the pair (e.g., <code>abc-xyz</code>) and that <code>r="RL"</code> or <code>"LR"</code> attributes are set up for the right direction.
+
#* '''Expand your morphology''' to cover '''at least 5 more elements''' of some paradigm(s).  This can be anything from additional verb or noun morphology, to adding all the forms of all the determiners (articles, demonstratives, etc.), to implementing nominal morphology on adjectives (e.g., if your language allows adjectives to be substantivised, which you'll want to add a tag for too).
#* Both <code>lrx</code> files are there and have the right names.
+
#* At least '''4 more twol rules''' that make your (analysis and) generation cleaner for an additional way.
#* All transfer files for both directions (up to 6 files) are there and have the right names and content.
+
#* At least '''3 new disambiguation rules''' that make the output of your tagger more accurate.
#* Also make sure that there are no compiled binaries or other compiled files committed to the repo.  If needed, use the <code>apertium-init</code> script to bootstrap a new pair to get the list of just the files that need to be in the repo, and use the the tricks presented in [[removing binaries from transducer repo]] to clean it up.
+
#* At least '''2 new lexical selection rules''' that make more of the right stems transfer over.
#* Make it clear on the pair's wiki page which repository link is the final resting place of all this (but leave a link to the other repo and don't remove it).  Also add a link to the final repo in the <code>README</code> in the superseded repo, with a note that the code has been merged into the other pair.
+
#* At least '''2 new transfer rules''' that make more of the output of your MT system closer to an acceptable target translation.
# '''Expand your MT pair''' in at least '''four''' of the following ways, listing in a "Final evaluation" section on the language pair's wiki page what you did (move existing evaluation sections under a new section called "Initial evaluation"), and for every rule (for all of the following except adding stems), list an example of what output was improved.
+
# When you are done with the above:
#* At least '''100 more stems''' in the bilingual dictionary (and monolingual dictionaries).
+
#* Document which of the above options you completed on the pair's wiki page (in a section like "Additions").  You don't have to list the words or rules you added, but do list that you added ''n'' words or ''n'' transfer rules or the like.
#* '''Expanded your morphology''' to cover '''at least 6 more elements''' of some paradigm(s).  This can be anything from additional verb or noun morphology, to adding all the forms of all the determiners (articles, demonstratives, etc.), to implementing nominal morphology on adjectives (e.g., if your language allows adjectives to be substantivised, which you'll want to add a tag for too).
+
#* Add a "polished RBMT system" tag to your repo.
#* At least '''4 more twol rules''' that make your (analysis and) generation cleaner.
+
#* '''document the following measures''' in the "Final evaluation" section of the pair's wiki page:.
#* At least '''4 new disambiguation rules''' that make the output of your tagger more accurate.
+
#** For your monolingual transducer:
#* At least '''3 new lexical selection rules''' that make more of the right stems transfer over.
+
#*** Updated precision and recall against the (updated) <code>eval.test</code> and <code>eval.gold</code> files (see [[Midterm overview#Evaluating your transducer]]),
#* At least '''3 new transfer rules''' that make more of the output of your MT system closer to an acceptable target translation.
+
#*** Coverage over the <code>large</code> corpus,
# When you are done with the above, '''document the following measures''' in the "Final evaluation" section of the pair's wiki page:.
+
#*** The number of words in the <code>large</code> corpus,
#* For each transducer:
+
#*** The number of stems in the transducer.
#** Precision and recall against the <code>annotated.basic</code> corpus,
+
#** For MT in the direction(s) you developed (<code>abc-xyz</code> and potentially <code>xyz-abc</code>):
#** Coverage over the <code>large</code> corpus,
+
#*** WER and PER over <code>longer</code> corpus.
#** The number of words in the <code>large</code> corpus,
+
#*** The proportion of stems translated correctly in the <code>longer</code> corpus.
#** The number of stems in the transducer.
+
#*** Trimmed coverage over <code>longer</code> and <code>large</code> corpora.
#* For MT in each direction:
+
#*** The number of tokens in <code>longer</code> and <code>large</code> corpora.
#** WER and PER over <code>longer</code> corpus.
 
#** The proportion of stems translated correctly in the <code>longer</code> corpus.
 
#** Trimmed coverage over <code>longer</code> and <code>large</code> corpora.
 
#** The number of stems in <code>longer</code> and <code>large</code> corpora.
 
  
 
[[Category:Assignments]]
 
[[Category:Assignments]]
 
[[Category:Tutorials]]
 
[[Category:Tutorials]]

Latest revision as of 09:09, 26 April 2022

If you're looking for information on corpus annotation and evaluation (precision/recall), that has moved to the Midterm overview page

Measuring trimmed coverage

Apertium RBMT systems use a "trimmed" version of the base transducer for each language in the MT pair. Each transducer is trimmed to only include entries found in the bilingual dictionary (.dix). The trimmed transducers are almost always smaller than the main transducers. Measuring coverage of the trimmed transducer gives you some idea of the ceiling of MT accuracy—or, at least, the percentage of forms in a corpus the MT system can analyse, regardless of translation accuracy.

Measuring trimmed coverage is just the same as measuring coverage, but with the appropriate "trimmed" transducer (e.g., xyz-abc.automorf.bin). You'll need to use the coverage-ltproc script instead of coverage-hfst.

The assignment

This assignment is due at the end of week 12 (this semester, at the end of the day on Friday, 29 April 2022, before midnight).

  1. Before you begin, make sure all previous assignments are done, and add a "structural_transfer" tag to your transducer repositories and your translation pair repository/ies to mark the end of previous assignments.
  2. Set up some new corpora based on existing ones:
    • Combine and merge your sentences and tests corpora so you [hopefully, but not necessarily] have a new longer parallel corpus. Name the files abc.longer.txt and xyz.longer.txt.
    • Make a large monolingual corpus of a bunch of raw text in your language. The more the better. This step may simply consist of you cleaning up and/or combining the existing corpora from the initial corpus assembly assignment. See if you can get it over 100K words. The bigger this corpus is the better. Call it abc.corpus.large.txt (in your monolingual corpus repo) and add notes to your MANIFEST file about where the text comes from.
  3. Expand your MT pair in at least three of the following ways for each translation direction that you're working on, listing in a "Final evaluation" section on the language pair's wiki page what you did (move existing evaluation sections under a new section called "Initial evaluation"), and for every rule (for all of the following except adding stems), list an example of what output was improved.
    • At least 100 more stems in the bilingual dictionary (and monolingual dictionaries as needed). This counts for both translation directions, if you are working on two-way translation. (If you've automated creation of your bilingual dictionary, then you can use this task to clean up 100 stems—at least 50 entries in your bilingual dictionary should be different than before—and properly categorise those 100 stems into the monolingual dictionary.)
    • Expand your morphology to cover at least 5 more elements of some paradigm(s). This can be anything from additional verb or noun morphology, to adding all the forms of all the determiners (articles, demonstratives, etc.), to implementing nominal morphology on adjectives (e.g., if your language allows adjectives to be substantivised, which you'll want to add a tag for too).
    • At least 4 more twol rules that make your (analysis and) generation cleaner for an additional way.
    • At least 3 new disambiguation rules that make the output of your tagger more accurate.
    • At least 2 new lexical selection rules that make more of the right stems transfer over.
    • At least 2 new transfer rules that make more of the output of your MT system closer to an acceptable target translation.
  4. When you are done with the above:
    • Document which of the above options you completed on the pair's wiki page (in a section like "Additions"). You don't have to list the words or rules you added, but do list that you added n words or n transfer rules or the like.
    • Add a "polished RBMT system" tag to your repo.
    • document the following measures in the "Final evaluation" section of the pair's wiki page:.
      • For your monolingual transducer:
        • Updated precision and recall against the (updated) eval.test and eval.gold files (see Midterm overview#Evaluating your transducer),
        • Coverage over the large corpus,
        • The number of words in the large corpus,
        • The number of stems in the transducer.
      • For MT in the direction(s) you developed (abc-xyz and potentially xyz-abc):
        • WER and PER over longer corpus.
        • The proportion of stems translated correctly in the longer corpus.
        • Trimmed coverage over longer and large corpora.
        • The number of tokens in longer and large corpora.