chrisyang
普通会员
CORPUSTOOL-ANNOUNCE-L] CorpusTool: Release 1.3 available
Michael O'Donnell <michael.odonnell@uam.es>
2008/2/5(周二), 下午5:42:21
[To stop receiving these emails, send me an
email to micko@wagsoft.com with subject: STOP CORPUSTOOL]
Dear All,
A new (beta) version of CorpusTool is on the website:
http://www.wagsoft.com/CorpusTool/download.html
Note that 'beta' indicates that the system may have bugs.
If you find any, let me know and I will fix them.
This version adds some new features, most related to searching:
1) Searches are now presented in a KWIK table format
(if you use the 'containing' keyword)
2) For 'x containing y' searches, you can now specify
'containing immediately', which is somewhat more
restrictive in its matchings. For instance, previously searching
for 'clause containing "Bush" ' would match both the
main clause and the dependent clause in:
[He said [that Bush was history]]
Now, this would only match the interior clause.
3) I added a very powerful feature, Autocoding, which
allows you to perform a search and then assign all matched
instances a particular feature from your scheme. For
instance,
Select PASSIVE if contains 'be% @participle'
...will present you with all instances of a be verb followed
by an -en verb, allow you to uncheck false matches, and
then code the remaining as passive.
Only more grammatical features can be coded in this way,
but it does allow the basic grammatical features to be coded
quite quickly. (ENGLISH only at this point)
Full release notes below:
====================================
Version 1.30 beta 23 January 2008
====================================
Major revision of Corpus Search, addition
of Autocode window (assign a feature to
a set of segments matching a search query).
SEARCH
* Major revision to the Corpus Search window.
- Now can use ^ in a search pattern to indicate
the pattern must match from the beginning of the segment.
- Added a keyword "containing immediately" in search queries
which means that where segments are embedded within others,
only the lowest level segment containing the search pattern
will match (e.g., "clauses containing '@modal-verb'" would
only match the embedded clause in [he said [she might go]] ).
- Now you can search for segments having one feature
but not others. Use the 'and not' conjunction when
extending the search query.
- Results displayed in a KWIK table if you
search for a segment CONTAINING other
segments, a string, or concordance pattern.
- The KWIK table can be sorted on any column
by clicking on the column header.
- The columns of the KWIK table can be dragged
wider or narrower.
AUTOCODING
* The main window now has a button "Autocode",
which opens an interface which allows you to define
autocode rules: rules such as:
select passive if contains 'be% @participle'
One selects one of these rules, and presses "Show",
and all instances in the corpus that match are presented.
You can then unselect any which are false matches (not
truly passives) and then click "Code". All selected
instances will be assigned the passive feature. Write
your own rules for the features you are interested in.
HELP
* This version has a revised online Help system. Each
window has a Help button, and clicking on it will show
the Help page for that topic.
CODING
* Reorganisation of the Coder Toolbar. There
is now a "Other Actions" selector which offers actions:
- "Export Annotations" alllows the annotation of this
file at this level is saved in an XML format,
with the annotations embedded within the
original text (CorpusTool annotation files
usually use standoff annotation, not containing the
original text, just char range pointers into
the original text).
- "Copy Features": remembers the feature so far coded for
the current segment.
- "Paste Features": assigns the features remembered with
Copu Features to the current segment.
STATS
* Bug fix: Comparisons of features between two sets,
results saved to file were wrong (in tstats and levels of significance)
although those shown on screen were correct (Thanks Thora).
Now fixed.
IMPORT LAYER
* There was a bug that stopped the "import coder study" option
(under the Project menu) from working. Now fixed.
Michael O'Donnell <michael.odonnell@uam.es>
2008/2/5(周二), 下午5:42:21
[To stop receiving these emails, send me an
email to micko@wagsoft.com with subject: STOP CORPUSTOOL]
Dear All,
A new (beta) version of CorpusTool is on the website:
http://www.wagsoft.com/CorpusTool/download.html
Note that 'beta' indicates that the system may have bugs.
If you find any, let me know and I will fix them.
This version adds some new features, most related to searching:
1) Searches are now presented in a KWIK table format
(if you use the 'containing' keyword)
2) For 'x containing y' searches, you can now specify
'containing immediately', which is somewhat more
restrictive in its matchings. For instance, previously searching
for 'clause containing "Bush" ' would match both the
main clause and the dependent clause in:
[He said [that Bush was history]]
Now, this would only match the interior clause.
3) I added a very powerful feature, Autocoding, which
allows you to perform a search and then assign all matched
instances a particular feature from your scheme. For
instance,
Select PASSIVE if contains 'be% @participle'
...will present you with all instances of a be verb followed
by an -en verb, allow you to uncheck false matches, and
then code the remaining as passive.
Only more grammatical features can be coded in this way,
but it does allow the basic grammatical features to be coded
quite quickly. (ENGLISH only at this point)
Full release notes below:
====================================
Version 1.30 beta 23 January 2008
====================================
Major revision of Corpus Search, addition
of Autocode window (assign a feature to
a set of segments matching a search query).
SEARCH
* Major revision to the Corpus Search window.
- Now can use ^ in a search pattern to indicate
the pattern must match from the beginning of the segment.
- Added a keyword "containing immediately" in search queries
which means that where segments are embedded within others,
only the lowest level segment containing the search pattern
will match (e.g., "clauses containing '@modal-verb'" would
only match the embedded clause in [he said [she might go]] ).
- Now you can search for segments having one feature
but not others. Use the 'and not' conjunction when
extending the search query.
- Results displayed in a KWIK table if you
search for a segment CONTAINING other
segments, a string, or concordance pattern.
- The KWIK table can be sorted on any column
by clicking on the column header.
- The columns of the KWIK table can be dragged
wider or narrower.
AUTOCODING
* The main window now has a button "Autocode",
which opens an interface which allows you to define
autocode rules: rules such as:
select passive if contains 'be% @participle'
One selects one of these rules, and presses "Show",
and all instances in the corpus that match are presented.
You can then unselect any which are false matches (not
truly passives) and then click "Code". All selected
instances will be assigned the passive feature. Write
your own rules for the features you are interested in.
HELP
* This version has a revised online Help system. Each
window has a Help button, and clicking on it will show
the Help page for that topic.
CODING
* Reorganisation of the Coder Toolbar. There
is now a "Other Actions" selector which offers actions:
- "Export Annotations" alllows the annotation of this
file at this level is saved in an XML format,
with the annotations embedded within the
original text (CorpusTool annotation files
usually use standoff annotation, not containing the
original text, just char range pointers into
the original text).
- "Copy Features": remembers the feature so far coded for
the current segment.
- "Paste Features": assigns the features remembered with
Copu Features to the current segment.
STATS
* Bug fix: Comparisons of features between two sets,
results saved to file were wrong (in tstats and levels of significance)
although those shown on screen were correct (Thanks Thora).
Now fixed.
IMPORT LAYER
* There was a bug that stopped the "import coder study" option
(under the Project menu) from working. Now fixed.
