lt-proc -d doesn't show exact (failing) input tags

The free and open-source rule-based machine translation platform

Brought to you by: jezral, mlforcada, nordfalk, sanmarf, and 3 others

#87 lt-proc -d doesn't show exact (failing) input tags

Status: open

Owner: nobody

Labels: lttoolbox (12) lt-proc (6)

Updated: 2017-05-18

Created: 2015-12-06

Creator: Kevin Brubeck Unhammer

Private: No

actual output:

$ echo '^omstende<n><nt><ut><sg><ind>$'|lt-proc -d nno/nno.autogen.bin 
  #omstende\<n\>\<nt\>\<sg\>\<ind\>

expected:

$ echo '^omstende<n><nt><ut><sg><ind>$'|lt-proc -d nno/nno.autogen.bin 
  #omstende\<n\>\<nt\>\<ut\>\<sg\>\<ind\>

where a correct analysis is e.g.

$ echo '^omstende<n><nt><sg><ind>$'|lt-proc -d nno/nno.autogen.bin 
omstende

1 Attachments

nno.autogen.bin

Discussion

Kevin Brubeck Unhammer - 2015-12-06

example fst

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Kevin Brubeck Unhammer - 2015-12-06

Attachments has changed:

Diff:

--- old +++ new @@ -0,0 +1 @@ +nno.autogen.bin (924.2 kB; application/octet-stream)
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Jim O'Regan - 2017-05-18

It looks like <ut> is just being deleted there:
$ echo '^omstende<n><ut><nt><sg><ind>$' |lt-proc -d /tmp/nno.autogen.bin </ind></sg></nt></ut></n></ut>

omstende\<n>\<nt>\<sg>\<ind>

echo '^omstende<n><ut><ut><sg><ind>$' |lt-proc -d /tmp/nno.autogen.bin </ind></sg></ut></ut></n>

omstende\<n>\<sg>\<ind>

$ echo '^omstende<n><ut><sg><ind>$' |lt-proc -d /tmp/nno.autogen.bin </ind></sg></ut></n>

omstende\<n>\<sg>\<ind>

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Jim O'Regan - 2017-05-18

echo '^omstende<n><ut><sg><ind><foo><bar>$' |lt-proc -d /tmp/nno.autogen.bin
#omstende\<n>\<sg>\<ind>
It's because there is no <ut> symbol in there[1]. Maybe retitle this to be that tags that aren't in the dictionary should be somehow marked?</ut></bar></foo></ind></sg></ut></n>

[1] lt-print /tmp/nno.autogen.bin |grep '<ut>' gives nothing, at least</ut>

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
- Kevin Brubeck Unhammer - 2017-05-19
  
  I was expecting the debug output to be exactly the input for the #-marked lu's. The string #omstende\<n>\<sg>\<ind> isn't anywhere (not in input, not in the bin), so it's rather unhelpful (and makes it hard to do testvoc using -d output).
  
  If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Log in to post a comment.