Recent changes to feature-requests

xml collocation window

2007-12-08T22:20:25Z

guy@sslmit.unibo.it

What one really wants this to do is to act not as an alternative to a word span, but in addition to a word span. In other words, if I select a word span of 4 left and 4 right, I want to be able to select <s> as an XML span which will limit my 4 words window to the same <s> as the node. So that a search for collocates of "day" in a span of left 4 right 4 in

<s>It 's a nice day is it ? </s><s>It was n't yesterday .</s>

will return two, not three occurrences of "it" as a collocate.

region not activated on startup

2007-12-08T22:01:10Z

guy@sslmit.unibo.it

Using BNCxml, having set preferences for region textonly, when I start up the client this region is displayed on the toolbar but doesn't apply when you do a query. To make it work you have to close the region and reselect it.

Thus a word query for Oxford give 16215 solutions on client startup, but 8815 once you cancel the region and then reselect it.

Additional statistics?

2007-01-07T11:40:02Z

Would it be possible to add other options for the collocation statistic? I am thinking particularly of log likelihood.

Grey out browse command

2006-12-28T01:58:11Z

When using the BNC across the Net, the "Browse" option in the "Texts" window is greyed out on the context (right-click) menu.

However, it isn't greyed out on the "Texts" menu, or on the context menu in a Query window, nor is the "Browse" button in the "Source" box disabled.

If possible it would be more user friendly to disable all these commands rather than rely on a dialog telling you it can't be done...

Activate class shortcut

2006-05-18T22:39:40Z

A suggestion for the user interface of the client: A
really useful extra feature would be an "Activate this
text's class" option or something like that on the
right-mouse-click option in the "Texts" window.

Copy word from Word Query list

2006-04-14T13:56:11Z

It would be very useful if there were a right-button-
click option to copy a word from a list in the Word
Query box -like the "Copy word" function in the
concordance view- particularly for when dealing with
alphabets you don't know very well (and thus would
find very painstaking to type).

Pattern in addkey

2006-03-08T23:06:07Z

Is it possible (or could it be made possible) to use
patterns in Addkey queries?

For instance, if you wanted to search for .*ed as JJ
as opposed to .*ed as VVD.

On the offchance that this was already possible, I
tried it in FLOB, and it seemed to *almost* work...
but when I pressed "refresh" I only got JJ on the tag
list, and when I tried to query this, I got no
solutions (<pos><word>.*ed</word><poscode
key="POS">JJ</poscode></pos>).

I have a vague memory of something like this being
raised on the preview list, but I can't recall what
the verdict was...

Andrew.

Save sort order as part of Word Query state

2006-03-05T12:52:30Z

Maybe this is a feature request rather than a bug, but
it would be very useful if the word query dialog could
remain in the most recent state (e.g. sorted by
frequency) when re-opened by a QueryEdit. It's quite
likely that one would want to see words in freq order,
popping off to look at KWIC queries for some of them
and then returning to the list.

Close Unicode keyboard via easy keystroke

2006-02-27T20:20:36Z

Recently I've been using Xaira lots fopr the analysis
of a Nepali corpus in whose ongoing development I am
involved, and have as a consequence been doing a lot
of typing in Devanagari.

It would be great if there were some easy way of
closing the Unicode keyboard from the physical
keyboard, e.g. if ENTER or ESCAPE closed the Unicode
keyboard and returned you to whatever bit of the Xaira
client you came from. I realise ALT+F4 does the job
but it's not an easy keystroke.

Although a small change it would make the interface
much easier to use, especially if you're doing lots of
queries in quick succession.

Allow transformation of labels in indexing

2005-07-26T11:06:59Z

Allow a more complicated transformation of label
attribute values when indexing. For example, with each
level of label the user would supply a pattern (default
(.*)) that would extract numbered variables from the
attribute value as in a regular expression. The user
would also supply a replacement string into which the
variables would be substituted: there would have to be
some way to distinguish variables arising from different
levels of label.

For example if I had paragraphs labelled PNNN and
sentences SMMM and I wanted the label text used to
be NNN/MMM I might supply:

paragraph pattern P(:d*)

sentence pattern S(:d*)

label text \1,1/\0,1

Are there examples eg in PERL or sed of replacement
patterns with multiple inputs? If so we should use their
syntax instead of \m,n.