Fwd: [WN-USERS] Sense Key Index
33 views
Skip to first unread message
german rigau
Jan 27, 2017, 10:44:13 AM1/27/17
to mcr-...@googlegroups.com, mc...@googlegroups.com, tuner-...@googlegroups.com
---------- Forwarded message ----------
From: Eric Kafe <ka...@megadoc.net>
Date: Fri, Jan 27, 2017 at 3:14 PM
Subject: [WN-USERS] Sense Key Index
To: WN-U...@princeton.edu
Hi all,
The Sense Key Index (SKI) now provides an easy and simple
pathway to interoperability between WordNet-related projects:
https://github.com/ekaf/ski
According to WordNet's "senseidx" manual page:
A sense_key is the best way to represent a sense
in semantic tagging or other systems that refer to
WordNet senses. sense_keys are independent of WordNet
sense numbers and synset_offsets, which vary between
versions of the database.
As a consequence, sense keys offer a stable basis for the
inter-operation between semantic web applications that rely on
different versions of WordNet.
Princeton WordNet (PWN) includes a sense key index
(the index.sense file) since version 1.3, but the implementation
of sense keys first became coherent in WordNet version 1.5.
Thus, we can define the full PWN sense key index as the unique
concatenation of all the coherent index.sense files from the
different versions of the original Princeton WordNet distribution.
The Sense Key Index can be used to:
- Generate database components in various formats (text, prolog, rdf),
to interface with any WordNet-related project: the GWA grid, OMW,
WN-ontology, ILI, MCR, Freeling, etc...
- Produce mappings between all WordNet versions
- Map version-bound WordNet resources like the ILI or the MCR
to other WordNet versions
- Produce statistics about the permanence of sense keys or ILI
identifiers across WordNet versions
- and more forthcoming...
This release of the SKI includes the following input databases:
- ski-pwn-sets.txt
sense key index for all the coherent Princeton WordNet versions
(currently 1.5, 1.6, 1.7, 1.7.1, 2.0, 2.1, 3.0 and 3.1)
- ski-mcr30-2016.txt
sense key index for MCR30-2016
- ski-ili30.txt
sense key index for ILI30
SKI-tools:
- Makefile
type "make all" to set executable permissions for the shell scripts
- pwn2maps
Generates synset offset mappings between all the WordNet versions
from ski-pwn-sets.txt
- pwn2flat
Generates the flat text relation between all synsets
in all WordNet versions and their sense keys.
Optionally, output this relation in Prolog format.
Additionally, produce a mapping from each sense key
to its last known WordNet version
- ili2map
(runs "pwn2flat" first, to generate the needed ski-pwn-flat.txt
and ski-pwn-last.txt databases)
Maps ILI-30 ids to all Princeton WordNet versions
Maps ILI-30 ids to their last known Princeton WordNet version
- mcr2free
Generates Freeling sense databases from MCR data
Output files:
For your convenience, this release includes all output from the SKI-tools,
compressed with gzip:
- ski-flat.tar.gz: the flat text version of ski-pwn-sets.txt
- ski-freeling.tar.gz: senses30.src databases for Freeling
- ski-ili.tar.gz: ILI mappings
- ski-mappings.tar.gz: PWN mappings
Please let me know whether the SKI is useful for your project.
Kind regards
Eric Kafe
.........................
HyperDic hyper-dictionary
http://www.hyperdic.net/
From: Eric Kafe <ka...@megadoc.net>
Date: Fri, Jan 27, 2017 at 3:14 PM
Subject: [WN-USERS] Sense Key Index
To: WN-U...@princeton.edu
Hi all,
The Sense Key Index (SKI) now provides an easy and simple
pathway to interoperability between WordNet-related projects:
https://github.com/ekaf/ski
According to WordNet's "senseidx" manual page:
A sense_key is the best way to represent a sense
in semantic tagging or other systems that refer to
WordNet senses. sense_keys are independent of WordNet
sense numbers and synset_offsets, which vary between
versions of the database.
As a consequence, sense keys offer a stable basis for the
inter-operation between semantic web applications that rely on
different versions of WordNet.
Princeton WordNet (PWN) includes a sense key index
(the index.sense file) since version 1.3, but the implementation
of sense keys first became coherent in WordNet version 1.5.
Thus, we can define the full PWN sense key index as the unique
concatenation of all the coherent index.sense files from the
different versions of the original Princeton WordNet distribution.
The Sense Key Index can be used to:
- Generate database components in various formats (text, prolog, rdf),
to interface with any WordNet-related project: the GWA grid, OMW,
WN-ontology, ILI, MCR, Freeling, etc...
- Produce mappings between all WordNet versions
- Map version-bound WordNet resources like the ILI or the MCR
to other WordNet versions
- Produce statistics about the permanence of sense keys or ILI
identifiers across WordNet versions
- and more forthcoming...
This release of the SKI includes the following input databases:
- ski-pwn-sets.txt
sense key index for all the coherent Princeton WordNet versions
(currently 1.5, 1.6, 1.7, 1.7.1, 2.0, 2.1, 3.0 and 3.1)
- ski-mcr30-2016.txt
sense key index for MCR30-2016
- ski-ili30.txt
sense key index for ILI30
SKI-tools:
- Makefile
type "make all" to set executable permissions for the shell scripts
- pwn2maps
Generates synset offset mappings between all the WordNet versions
from ski-pwn-sets.txt
- pwn2flat
Generates the flat text relation between all synsets
in all WordNet versions and their sense keys.
Optionally, output this relation in Prolog format.
Additionally, produce a mapping from each sense key
to its last known WordNet version
- ili2map
(runs "pwn2flat" first, to generate the needed ski-pwn-flat.txt
and ski-pwn-last.txt databases)
Maps ILI-30 ids to all Princeton WordNet versions
Maps ILI-30 ids to their last known Princeton WordNet version
- mcr2free
Generates Freeling sense databases from MCR data
Output files:
For your convenience, this release includes all output from the SKI-tools,
compressed with gzip:
- ski-flat.tar.gz: the flat text version of ski-pwn-sets.txt
- ski-freeling.tar.gz: senses30.src databases for Freeling
- ski-ili.tar.gz: ILI mappings
- ski-mappings.tar.gz: PWN mappings
Please let me know whether the SKI is useful for your project.
Kind regards
Eric Kafe
.........................
HyperDic hyper-dictionary
http://www.hyperdic.net/
german rigau
Jan 30, 2017, 4:05:00 PM1/30/17
to Xavier Gómez Guinovart, tuner-...@googlegroups.com, mc...@googlegroups.com, mcr-...@googlegroups.com
Buena pregunta ... como estoy en Tartu con Francis, le pregunto.
:-)
German
On 30 Jan 2017 21:06, "Xavier Gómez Guinovart" <x...@uvigo.es> wrote:
Hola German
Después de ejecutar el script ili2map del paquete, y de obtener una salida con líneas como las de abajo, nos queda una duda. El identificador inicial de cada línea tiene algo que ver con el CILI? Están relacionados de algún modo el SKI de Eric Kafe y el CILI?:
...
i45045 2.0:n:01765628
i45045 2.1:n:01827090
i45045 3.0:n:01846331
i45045 3.1:n:01848972
...
Saludos,
Xavier
----
You received this message because you are subscribed to the Google Groups "tuner-project" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tuner-project+unsubscribe@googlegroups.com.
To post to this group, send email to tuner-...@googlegroups.com.
Visit this group at https://groups.google.com/group/tuner-project.
To view this discussion on the web visit https://groups.google.com/d/msgid/tuner-project/CAB00nwoBP%3Dk2wP366VsiUqYiTt7negX4QZ194_WwPNrRGVd5jA%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.
You received this message because you are subscribed to the Google Groups "tuner-project" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tuner-project+unsubscribe@googlegroups.com.
To post to this group, send email to tuner-...@googlegroups.com.
Visit this group at https://groups.google.com/group/tuner-project.
To view this discussion on the web visit https://groups.google.com/d/msgid/tuner-project/f94786f2-c2a8-d462-27aa-453677d833da%40uvigo.es.
For more options, visit https://groups.google.com/d/optout.
Eric Kafe
Feb 10, 2017, 5:59:56 AM2/10/17
to MCR-users, x...@uvigo.es, tuner-...@googlegroups.com, mc...@googlegroups.com
Hi
Regards
Eric
The updated README file of the SKI package at
https://github.com/ekaf/ski clarifies the description
of the input databases, in particular: - ski-mcr30-2016.txt sense key index for MCR30-2016, derived by joining the inverse SKI (ski-pwn-iflat.txt) with the latest MCR "variant" files, retrieved from http://adimen.si.ehu.es/web/MCR - ski-ili30.txt sense key index for ILI30, derived by joining the inverse SKI (ski-pwn-iflat.txt) with the GWA-ILI "ili-map-pwn30.tab" file retrieved from https://github.com/globalwordnet/ili
Regards
Eric
To unsubscribe from this group and stop receiving emails from it, send an email to tuner-projec...@googlegroups.com.
To post to this group, send email to tuner-...@googlegroups.com.
Visit this group at https://groups.google.com/group/tuner-project.
To view this discussion on the web visit https://groups.google.com/d/msgid/tuner-project/CAB00nwoBP%3Dk2wP366VsiUqYiTt7negX4QZ194_WwPNrRGVd5jA%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.
--
You received this message because you are subscribed to the Google Groups "tuner-project" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tuner-projec...@googlegroups.com.
Reply all
Reply to author
Forward
0 new messages