[CWB] Public Frequency Lists

[Hardie, Andrew]

Hi Simon,

No, it's not the case that every corpus shows up - or at least, not for everyone; what you see depends on which user you are. *You* are seeing all corpora incl. invisible because, I guess, you are superuser.

Here's the new system (as of 3.2.40 I think) in detail.

Currently, in both 3.2 and 3.3, the list of corpora/subcorpora offered as reference for keywords includes the following:

    *       - Remainder                               (current corpus minus subcorpus selected on list 1)
    *       - Subcorpora (local, owned)               (lus)
    *       - Granted subcorpora (local)              (lgs)
    *       - Public subcorpora (local, non-owned)    (lps)
    *       - Entire corpus                           (***)
    *       - System corpora (any access level)       (fsc)
    *       - User's own corpora.                     (fuc)
    *       - Granted corpora.                        (fgc)
    *       - User's nonlocal subcorpora              (fus)
    *       - Public subcorpora (nonlocal, nonowned)  (fps)
    *       - Granted subcorpora (nonlocal, nonowned) (fgs)

where "local" = part of the current corpus ; "owned" = installed by the user in the "user corpora" system (overall still buggy! but this bit works I think) ; "granted" = installed by another user who has given the current user access; summarised by the three-letter code. (Above is from a comment in query-forms.php circa line 840ff in 3.3 and circa 940 ff. in 3.2) [*]

That is, as well as public subcorpora, users are able to use the freq list of any system or user corpus they have a suitable privilege/grant for. There's no such thing as a public corpus FL any more (and both the database field and UI to set it are gone), only corpora that the user has permission for. 

This is part of a long-planned move away from the ancient all-or-nothing "public freqlist" mechanism (which was always a bit of a bodge) to relying on the "corpus access privilege" system, which has at least some semblance of flexibility. "Public FL" still exists for subcorpora, though.

In future, the public FL feature will be removed completely once I work out how to handle subcorpora with the "permissions" system.  

Other plans: a new privilege level, perhaps, IE "freqlist only access" or similar, which would grant a level of access similar to the previous "public" function. Also in future, the list of corpora available for KW will be filtered by language *iff* the language of the local corpus is not "undetermined". 

Till then, subcorpus FL publicness is controlled just as it always was: through "cached frequency lists". 

best

Andrew.

[*] see https://sourceforge.net/p/cwb/code/HEAD/tree/gui/cqpweb/branches/3.2-latest/lib/query-forms.php#l931 


-----Original Message-----
From: cwb-bounces at sslmit.unibo.it <cwb-bounces at sslmit.unibo.it> On Behalf Of Simon Meier-Vieracker
Sent: 08 October 2020 15:46
To: Open source development of the Corpus WorkBench <cwb at sslmit.unibo.it>
Subject: [CWB] Public Frequency Lists


Hi,

in older versions of CQPweb (e.g. v3.2.27) there was an option under „Manage frequency lists“ to provide public frequency lists that could be used for keyword calculation.

Am I right, that In the current version 3.2.42 every corpus, even the invisible ones, show up as frequency lists in the keywords section by default? I can’t find the button to make frequency lists private, but I would like to do so...

Best, Simon