IRC log for #koha, 2006-03-06

← Previous day | Today | Next day → | Search | Index

All times shown according to UTC.

Time Nick Message
15:40 kados paul_away: you don't happen to be present do you?
16:23 thd: you there?
16:23 thd: I'm looking at the ISBD display for authorities currently
16:24 thd kados: good morning :)
16:24 kados I think what we want to be able to do is something like this:
16:24 <heading>
16:24 <auth_heading>[100a] [100d]</auth_heading>
16:24 <see>[400a] [400d]</see>
16:24 <see_also>[500a] [500d]</see_also>
16:24 </heading>
16:24 (for the markup)
16:24 (good morning :-))
16:24 each <see> and <see_also> tag is repeatable
16:25 thd kados: where is that code?
16:26 kados it doesn't exist yet
16:26 thd :)
16:26 kados I spent a frustrating morning trying to figure out how the headings _should_ display
16:26 thd kados: where is the place where it goes?
16:26 kados uthoritiesMarc.pm
16:26 AuthoritiesMarc.pm even
16:26 wait ... no ... the xml you see above would go in the 'summary'
16:27 thd kados: what is there currently that serves that function?
16:27 kados another alternative is to just have the system be smart enough to generate the summary from the data
16:27 that's what you wanted I think
16:28 thd by summary do you mean authorised heading?
16:29 kados: remember you changed the column name from summary in at least one place
16:29 kados yep
16:30 but the variable is still called $summary and summary
16:30 anyway, I think I have an idea for how to do it
16:30 since the 'See' is always of form 4XX and 'See also' is always of form '5XX'
16:31 thd kados: I think that variable was meant to hold the authority framework type summary
16:31 kados headings are always of the form 1XX
16:32 yep it was
16:32 but that has problems
16:32 it can't deal with ordering
16:32 thd kados: maybe it is too much code to change for the moment but identifying the framework type if searching for more than one framework type would be helpful
16:32 kados and repeatability
16:32 give me 5 minutes and I'll have something impressive :-)
16:33 thd kados: are you intending to display the 4XX and 5XX returned from the authority?
16:33 kados yep
16:34 thd kados: If you had real authorities would that not make a crowded display
16:35 kados: why would you not display only the 1XX or 7XX (where displaying 7XX is dependent upon the context being past $a)
16:36 kados: also I have been up earlier than I should have looking at the current display
16:38 kados: the authorised form that appears in the matching bibliographic record is what should be a hyperlink.
16:39 kados: The biblio count should just be a count with no link embedded.
16:40 kados: The additional link with the authority record number or set of record numbers would be another link to the MARC record for the authority or even an authority detail display.
16:42 kados thd: I was using loc as a model
16:43 thd kados: Existing systems link the authorised heading not the biblio count to the biblios.  The authorities.loc.gov representation is unusual, not extensively developed, and fortunately not widely used.
16:43 kados k ... that's easy enough to change
16:44 so the heading should display 1XX entries ... or else 7XX entries ... right?
16:44 no See or See also entries?
16:44 thd kados: LC is using left anchored searches only eeeewwww :0
16:44 kados thd: if we could reproduce the functionanlity there I would be happy as a clam :-)
16:44 thd: though I agree it could be better
16:45 thd kados: see also would be useful if it led to a different authority
16:45 kados http://www.loc.gov/marc/uma/pt1-7.html#pt4
16:45 in that document (maybe the next page)
16:45 thd kados: see also is 5XX I think
16:46 kados they describe display of See and See also entries for opac interfaces
16:48 thd kados: so is Clements, Samuel Langhorne, 1835-1910 a different authority record from Twain, Mark ?
16:49 kados: that display is stuck in the printed card age
16:50 kados thd: look at the name auth record for Lewis now
16:51 thd kados: although that would make it explicit to the user why searching for Samuel Clements found Mark Twain
16:51 kados: I take back what I said about the printed card age and crowded displays
16:52 kados: making the connection to the search term used by the user to arrive at the result is useful
16:52 s/connection/connection explicit/
16:54 kados: I have seen Lewis now but it does not match the LC model
16:54 kados it is missing some lines
16:54 actually the see is in the wrong place
16:55 kados thd: you sure?
16:55 thd kados: nevemind
16:56 kados: those are different people
16:57 kados: no they are the same person are the not?
16:58 kados: I had it right originally you need extra lines
16:58 kados: you now have ...
16:59 Lewis, C. S.1898-1963 (Clive Staples),
16:59 See Lewis, Jack,1898-1963
16:59 See Hamilton, Clive,
16:59 kados: you need to add
17:02 the equivalent of ...
17:02   Clemens, Samuel Langhorne, 1835-1910
17:02      see also: Twain, Mark, 1835-1910
17:02   Conte, Louis de, 1835-1910
17:02      see: Twain, Mark, 1835-1910
17:02 kados: underneath what you have already
17:04 kados I did
17:05 thd kados: otherwise the user thinks that searching under a different 4XX term is going to find different set of bib hits
17:05 kados: let me explain the difference
17:06 kados how is what you posted above different from:
17:06 Lewis, C. S.1898-1963 (Clive Staples),
17:06 see: Lewis, Jack,1898-1963
17:06 see: Hamilton, Clive,
17:06 thd you have only the equivalent of .
17:06 Twain, Mark, 1835-1910
17:06      see also: Clements, Samuel Langhorne, 1835-1910
17:07 kados what I last posted comes before what I had first posted
17:07 for the LC example.
17:08 kados ahh ... so both the authorized and unauthorized should show up ...
17:08 is that what you mean?
17:08 thd kados: if all the user sees is see Lewis, Jack 1898-1963 then
17:09 kados thd: if you search 'anywhere' for Lewis, Jack
17:09 thd: it will still return the correct authorized heading
17:09 thd the user may expect that searching on Lewis Jack you would get different bib hits
17:10 kados thd: the user will never see Lewis, jack as a main heading result
17:10 with the current scheme
17:10 thd kados: exactly but you have not made that clear to the user as in the LC example
17:10 kados so maybe we should have 'highlighting' so they can see where it matches
17:10 what do you think?
17:11 thd kados: look at the LC example for every see they also have the line showing that you get right back to the record you have already
17:11 kados thd: we have no results for 'see' that I know of
17:12 thd: the user is always sent to the authority record
17:12 thd kados: I am doing a poor job of explaining this
17:12 kados: I will change the LC example to match the extra lines needed for Lewis
17:13 kados: I did not want you to wait while I typed them
17:13 kados: I will do 2 lines at a time
17:14 kados ok
17:14 thd kados: this much is fine for now although indentation and authorised term in bold would help
17:15 Lewis, C. S.1898-1963 (Clive Staples),
17:15 see: Lewis, Jack,1898-1963
17:15 see: Hamilton, Clive,
17:15 kados that's what I have already
17:20 thd kados: it should be as follows
17:20 Lewis, C. S.1898-1963 (Clive Staples),
17:20 see: Lewis, Jack,1898-1963
17:20 see: Hamilton, Clive,
17:20 oops
17:21 kados starting again  ...
17:21 Lewis, C. S.(Clive Staples), 1898-1963.
17:21       see: Lewis, Jack,1898-1963
17:21       see: Hamilton, Clive,
17:21   Lewis, Jack,1898-1963
17:21      see also: Lewis, C. S. (Clive Staples), 1898-1963.
17:21   Hamilton, Clive,
17:21      see also: Lewis, C. S. (Clive Staples), 1898-1963.
17:21 kados: note the change in date ordering
17:27 kados thd: try searching now for Hamilton, Clark
17:27 thd: or Lewis, Jack
17:28 thd: it will now style the terms you used in the results
17:28 thd kados: #100||{ 100a }{ 100b }{ 100c }{ 100q }{ 100d }{ 100e}|
17:29 kados thd: tell me what you think of my change
17:29 thd: what did you post?
17:31 thd kados: I pasted just now the correct place for X00q and also spacing separating the subfields
17:31 i.e. the correct relative place for $q
17:32 kados thd: currently, the display is ordered in the same way as the record
17:32 thd: (in my modified display that is)
17:32 thd kados: the record is wrong :)
17:32 kados thd: ok :-)
17:32 thd: that's a record editor problem
17:33 thd: if a correct record were imported it would display correctly
17:33 thd kados: oh yes you have not fixed the authority record editor yet
17:34 kados: I understood that searching on a 4XX, 5XX, etc would let me find the authority for 1XX
17:35 kados it does
17:35 thd kados: do you not see the added value of 4 more lines?
17:35 kados four more lines?
17:35 which lines?
17:35 thd   Lewis, Jack,1898-1963
17:35      see also: Lewis, C. S. (Clive Staples), 1898-1963.
17:35   Hamilton, Clive,
17:35      see also: Lewis, C. S. (Clive Staples), 1898-1963.
17:36 kados: maybe they should be see not see also
17:36 kados I don't understand
17:36 a search for Lewis gives me:
17:36 Lewis, C. S.1898-1963 (Clive Staples),
17:36   see: Lewis, Jack,1898-1963
17:36   see: Hamilton, Clive,
17:37 why would you want Lewis Jack and Hamilton Clive listed separately from the main authority?
17:37 especially now that they are highlighted in red if you searched for them
17:37 thd kados: that much conveys to the user that if you search for Lewis, Jack you will find something different
17:37 kados but you wont!
17:38 thd kados: the extra lines qualify that
17:38 kados in fact, they are all part of the same authority record
17:38 thd kados: of course so make that explicit
17:38 to the user
17:38 the extra lines do that
17:38 kados wow ... I really don't understand
17:38 Lewis, C. S.1898-1963 (Clive Staples),
17:38   see: Lewis, Jack,1898-1963
17:38 thd :0
17:38 kados   see: Hamilton, Clive,
17:38 that's what I have now
17:38 thd yes
17:39 kados: that looks like an instruction to the user to me
17:39 kados you want those 'see' entries listed separately _again_?
17:39 thd: but that's just the stupidity of MARC
17:39 thd signifying go and search for Lewis Jack to find something
17:40 kados thd: it's not my fault they use 'see' and 'see also'
17:40 thd kados: which is why user displays have to be more informative
17:40 kados: or less informative as I had wrongly suggested originally
17:41 kados so in marc authorities
17:41 you've got one main authorized heading in the 1XX
17:41 then you've got alternative authorized headings in the 4XX
17:41 then you've not unauthorized headings in 5XX
17:41 s/not/got/
17:42 thd kados: right underneath there is the qualification that searching otherwise will find where you are already
17:42 kados what you want is for a separate entry for all of the headings
17:42 thd kados: the see is actually a see from
17:43 kados that links to the main authorized heading
17:44 thd kados the see in the authority is supposed to be understood as a see from in the authorised authority
17:44 kados: So as this was derived from a card catalogue
17:45 kados I think I understand
17:45 I'll see what I can do
17:45 thd kados: there would have been other cards matching the additional lines
17:46 kados: the main card would not have had the see from as I recall
17:47 kados: Card catalogues are completely gone but you can still see this in the massive printed volumes of LCSH
17:47 kados: I have 2 copies of those massive volumes
17:48 kados: unfortunately they take up too much space for me to keep them at hand :)
17:52 kados it's quite hard in fact
17:53 because as the object comes out there is know way to know what the last subfield iw
17:53 is for a given tag
17:55 I may need to go back a level and take a look at how this object is created
17:57 I know how to do it
17:57 I'll just hand the different types to the template
17:57 in a loop
18:29 thd: man, you're not kidding isbd was done backwards!
18:29 thd: i can't even begin to figure out how to fix it
18:29 thd kados: I had fixed what you had originally
18:30 kados thd: er?
18:30 thd: where?
18:30 thd <heading>
18:30    <auth_heading><b>[ 100a ] [ 100b ] [ 100c ] [ 100q ] [ 100d ] [ 100e ]</b></auth_heading>
18:30    <see>[ 400a ] [ 400b ] [ 400c ] [ 400q ] [ 400d ] [ 400e ]</see>
18:30    <see_also>[ 500a ] [ 500b ] [ 500c ] [ 500q ] [ 500d ] [ 500e ]</see_also>
18:30 </heading>
18:30 kados thd: I had to revert back to cvs code
18:30 thd: because what i had was too convoluted
18:31 thd kados: the <b> belongs in the template really
18:31 kados thd: here's the problem with isbd
18:31 thd kados: I have a simpler solution
18:31 kados if you put [ 100a ] [ 100b ] [ 100c ]
18:31 and in your data you have two of each of those (repeatable)
18:31 they will come out like this :
18:32 [ 100a ] [ 100a ] [ 100a ] [ 100b ] [ 100b ] [ 100b ] [ 100c ] [ 100c ] [ 100c ]
18:32 thd kados: but I do not know the template module that Koha uses yet
18:32 kados instead of
18:32 [ 100a ] [ 100b ] [ 100c ] [ 100a ] [ 100b ] [ 100c ] [ 100a ] [ 100b ] [ 100c ]
18:32 thd: HTML::Template
18:33 thd kados: I mean that I have not learnt the usage for the module yet
18:33 kados ahh
18:33 I'm going back a level to look at how the fields are coming out from the db
18:34 thd kados: I have used the CGI module for the same purpose in the past
18:34 kados aak!
18:34 my @fields = $record->fields();
18:34 wtf
18:35 that's an insane way to deal with MARc data
18:35 thd kados: yes repeated fields work improperly unless they happen to be repeated just next to one another in the original record
18:36 kados: It is a simple way if all the records that you looked at led you to believe that orderliness matched alphabetic order
18:37 kados: Which is what paul had presumed from the data he had seen
18:38 kados: yet he seemed to have even been using examples from libraries that were not following the UNIMARC standard correctly already.
18:39 kados: Getting it correct requires looking hard for the most difficult records.
18:39 kados: If you can do the problematic ones then you can do any ones
18:42 kados in my example remove the spaces between ] and [ to get ..
18:43    <auth_heading><b>[ 100a ][ 100b ][ 100c ][ 100q ][ 100d ][ 100e ]</b></auth_heading>
18:43    <see>[ 400a ][ 400b ][ 400c ][ 400q ][ 400d ][ 400e ]</see>
18:43    <see_also>[ 500a ][ 500b ][ 500c ][ 500q ][ 500d ][ 500e ]</see_also>
18:44 kados: although spaces at that point make no difference
18:45 kados: spaces need to be inside the brackets to function
18:54 kados spaces don't do anything that I can tell
18:54 the major problem here
18:55 is that there is no acknoledgement that this is a hierarchy
18:55 with subfields being 'inside' of tags
18:56 thd kados: spaces work inside curly braces for the ISBD preference
18:56 kados curly braces?
18:57 thd { 100a }
18:58 multiple spaces are reduced to one by browser rendering so that would be no problem
18:59 kados i don't think ISBD is meant to be user configuratble
18:59 thd kados: what lack of acknowledgement are you referencing.  Do you mean the user display?
19:00 kados it seems from the spec like it's just meant to be pre-defined
19:00 thd: yes from the user display
19:01 thd kados the user display can use indentation with an HTML dictionary list
19:01 kados I have no idea what you mean by that
19:01 thd kados: dictionary usually render hierarchically but that can be enforced in CSS
19:02 kados <dl>
19:04 kados:  <dd> will give you a drop indent
19:13 kados there is no way we can represent things in the right order using the current code
19:14 thd kados: I supplied you with ISBD code that forces things to the order that is usually correct
19:14 when I substituted your example XML
19:15 kados: The correct was is the routine from SearchMarc.pm
19:15 kados: the one you looked at last night
19:16 s/was/way/
19:20 kados: paul had created getMARCsubjects in SearchMarc.pm for this very problem.
19:20 kados: I understand that code well and can adapt it for any field
19:22 kados: It is really very simple after staring at it for long enough :)
19:24 kados: It needs to be generalised with a variable for what type of field it needs to return so it knows how to apply spaces and/or '---' between subfields.
19:26 Where the '--' does not apply to name fields but only to topic fields.
19:34 kados thd: check the Lewis entry now
19:34 thd: see if it's what you want
19:35 thd: it should work properly if the data is correct
19:35 thd: of course, it's not the programmers reponsibility to fix improper cataloging
19:35 thd kados: you can force it to appear more correctly
19:36 kados (unless it's caused by improper MARc edit tools :-))
19:36 thd: that is a version 4 goal :-)
19:36 thd: I would be happy currently to just get it working with correct data
19:36 thd: how does the Lewis entry look to you?
19:37 thd kados: if you are using the existing Koha ISBD backwards system then you have forced it to be incorrect even if the record were correct
19:37 kados thd: I'm not
19:37 thd: I rewrote the way that $summary is populated
19:37 thd kados: However, you can easily fix that within the broken ISBD system
19:38 kados the current ISBD system must die
19:38 it can't represent a hierarchy
19:38 thd kados: using code like what is in getMARCsubjects ?
19:38 kados so in my mind it's never going to work properly
19:38 I'll have to reread that code
19:38 but if I recall there is much cardcoded in it
19:39 IMO this is a better approach:
19:39 foreach my $field ($record->field('1..')) {
19:39                        $heading.= $field->as_string();
19:39                }
19:39                my $summary.="<b>".$heading."</b><br>";
19:39                foreach my $field ($record->field('4..')) {
19:39                        $summary.= $field->as_string()."<br>";
19:39 thd kados: That is why you need to pass a couple of variables to generalise it
19:39 kados                        $summary.= "&nbsp;&nbsp;&nbsp;<i>see:</i> ".$heading."<br>";
19:39                }
19:41 thd kados: that works although it does not match the LC example exactly yet.
19:41 kados: I can fix it though
19:41 kados thd: only because the data is incorrect :-)
19:42 thd kados: I mean indentation boldness and the two see from lines just under the heading.
19:43 kados: but what you have is not confusing and looks beautiful
19:43 kados thd: it matches this perfectly I think:
19:43 http://www.loc.gov/marc/uma/pt12.html
19:44 thd sorry you have boldness, and partial indentation
19:44 kados so do they :-)
19:45 boldness for the first heading
19:45 thd kados: you have the last 4 lines but you dropped the 2nd and 3rd line
19:45 kados: and if you look closely everything after the heading is indented relative to the heading
19:46 kados Lewis, C. S. 1898-1963 (Clive Staples),
19:46 Lewis, Jack, 1898-1963
19:46   see: Lewis, C. S. 1898-1963 (Clive Staples),
19:46 Hamilton, Clive,
19:46   see: Lewis, C. S. 1898-1963 (Clive Staples),
19:46 what is the difference?:
19:46 Woolf, Virginia, 1882-1941
19:46 Stephen, Virginia, 1882-1941
19:46 See Woolf, Virginia, 1882-1941
19:46 Woolf, Virginia Stephen, 1882-1941
19:46 See Woolf, Virginia, 1882-1941
19:47 thd compare ...
19:47 Lewis, C. S.(Clive Staples), 1898-1963.
19:47       see: Lewis, Jack,1898-1963
19:47       see: Hamilton, Clive,
19:47   Lewis, Jack,1898-1963
19:47      see also: Lewis, C. S. (Clive Staples), 1898-1963.
19:47   Hamilton, Clive,
19:47      see also: Lewis, C. S. (Clive Staples), 1898-1963.
19:47 kados: i have confused see and see also so ignore that
19:48 kados ok
19:48 thd kados: look at the indentation
19:48 kados but actually you can't
19:48 we must be looking at different pages
19:48 thd kados can't what?
19:48 kados see also means it's authoritative
19:48 see means it's not
19:49 so you can't ignore that because I think it should affect the display
19:49 thd that was my mistake
19:49 kados: I just had not corrected that when I posted that again
19:49 assume I had posted that bit correctly
19:50 kados: just look at the relative indentation
19:50 Lewis, C. S.(Clive Staples), 1898-1963.
19:50       see: Lewis, Jack,1898-1963
19:50       see: Hamilton, Clive,
19:50   Lewis, Jack,1898-1963
19:50      see: Lewis, C. S. (Clive Staples), 1898-1963.
19:50   Hamilton, Clive,
19:50      see: Lewis, C. S. (Clive Staples), 1898-1963.
19:51 kados ok
19:51 thd kados: now that it is fixed focus on how everything is indented relative to the heading
19:51 kados: also they have lines 2 and 3
19:52 kados there shouldn't be 'see' options under the main heading
19:52 thd kados: I do not understand why they have the see from listing in their example but they do
19:53 in lines 2 and 3
19:54 kados: we were referencing different pages just now
19:55 kados: I had modelled your first Twain example from earlier
19:55 kados right, I see that now
19:55 I'm changing it ... just a sec
19:56 thd kados: I do not know what context is different except that I suspect the earlier example was for a computer display
19:59 kados; well they are both obviously for computer displays but the second one looks more like a sequential browse in a large alphabetic list
20:03 kados: I have never actually read the Understanding MARC Authority Records.  It is not as old as Understanding MARC Bibliographic which first helped me to understand MARC well.
20:04 kados thd: in the Lewis example
20:04 thd: there are no alternative authorized headings
20:04 thd: only unauthorized ones
20:04 thd: (on liblime's opac)
20:05 unauthorized headings reference authorized headings
20:05 authorized headings reference other authorized headings
20:06 but authorized headings never reference unauthorized headings
20:06 thd kados: so that is the difference or the degree of indentation?
20:06 kados yep
20:06 we need an example with multiple authorized headings to test my code
20:07 thd kados: no they have both
20:07 kados: the Twain example has both 500 and 400
20:07 kados thd: but that's not on LibLime's opac is it?
20:08 thd kados: yet the 400 is still indented underneath th heading
20:08 kados it's indented
20:09 right
20:09 but there is no 'see' entry for it
20:09 thd kados: exactly
20:10 Lewis, C. S.(Clive Staples), 1898-1963.
20:10   Lewis, Jack,1898-1963
20:10      see: Lewis, C. S. (Clive Staples), 1898-1963.
20:10   Hamilton, Clive,
20:10      see: Lewis, C. S. (Clive Staples), 1898-1963.
20:11 kados: I think if you outdent the 4XX then that is for browsing a large alphabetical list
20:12 kados: This indentation I just posted seems more appropriate.
20:13 kados thd: bingo
20:13 thd: look now
20:13 thd kados: the IFLA standards for OPAC displays have been published for a few months but I have not checked to see if it has hit the web yet
20:14 kados we still need to test cases where there are 'see also' entries
20:15 thd kados: looks good, now move the hyperlink from the number of bib records to the heading
20:15 kados hehe ... trickier than it seems :-)
20:15 lets test the seealso case first
20:15 thd kados: ok
20:16 kados: we know that we can create Twain
20:17 kados adding it now
20:18 hehe ... 500 error :-)
20:20 thd kados: really, I have no 500 error
20:20 kados: which intranet subdomain is this?
20:29 IFLA Guidelines for Online Public Access Catalogue (OPAC) Displays.
20:29 Final Report May 2005
20:29 M?nchen: Saur, 2005, 61 p.
20:29 ISBN 3-598-24276-X
20:29 (IFLA Series on Bibliographic Control; vol. 27).
20:29 Price: EUR 34 (IFLA Members EUR 26,80)
20:31 kados: it does not seem to be on the web yet although some inadequate early draft is there from before politics and time wore down the draft author.
20:40 kados thd: check the opac for 'Twain' now
20:40 we may have achieved perfection :-)
20:40 thd kados: i just checked the intranet and it is off just a little with spacing :)
20:41 kados heh
20:41 check again
20:42 thd the OPAC is fine
20:42 the intranet is the same :)
20:43 I mean same fine :)
20:43 kados great
20:44 so what's next? :-)
20:45 thd kados: a dictionary list with the right CSS might look just slightly better typographically than break tags
20:45 however once everything else is perfect that could be fiddled with
20:46 kados yep
20:46 thd kados: what about linking the heading instead of the hit count?
20:46 kados I'll try
20:48 thd kados: I know that everything throughout Koha mostly tries to use the anchor link content identical to the anchor display content.
20:53 kados woot ... working
20:57 thd: you sure the heading shouldn't take you to the MARC?
20:57 thd wow no not in the usual displays that I have seen
20:58 kados: and for the MARC in the OPAC that should be on a rightmost column
20:59 kados: the patrons want to find the material catalogued not usually the MARC displays
20:59 kados what should be on the right-hand column?
21:00 thd kados: in the OPAC at least the link to the MARC record for the authority should be a last column, not a first column
21:01 kados gotcha
21:04 ok ...
21:04 subject headings don't really work very well
21:04 i suspect that's because the bulk import didn't do everything
21:04 thd kados: Authorized Heading(#4074) could use better display text such as MARC for whatever is in the 1XX
21:05 kados give me an example
21:06 thd kados: MARC for Twain, Mark, 1835-1910 but I am sure the description could improve with some consideration and examination of other systems.
21:07 kados: and I hope that delete does nothing in the OPAC :)
21:08 kados heh ... aactually it might actually delete :-)
21:08 thd delete should not be in the OPAC :)
21:08 kados I'll need to get rid of that :-)
21:09 thd kados but  you left the column in heading row
21:10 kados yea i intend to put the type in there
21:10 thd beautiful
21:10 kados personal name, corporate name, etc.
21:11 thd The OPAC should still have nicer text instead of Authorized Heading(#4074)
21:11 kados like what?
21:13 thd MARC for Twain, Mark, 1835-1910 or something better after more thought and example examination from other systems
21:13 certainly patrons do not care what the record number is
21:15 kados how's that?
21:15 thd kados: much better
21:15 kados: that is beautiful
21:18 kados how do we tell what type of heading it is most efficiently?
21:18 thd kados: we have to overcome the problem for building records whee it did not find the records
21:19 kados there is much to do in building a given record yet
21:19 but I'm happy we're getting there :-)
21:19 thd kados: my suggestion was to use only 'a' as the key
21:20 kados hmmm ... I still don't fully understand what paul's hash is used for
21:20 thd kados: yes, I am merely tossing out a guess without real inspection of the code or asking paul or hdl
21:21 kados I suspect we will need to rewrite it
21:21 the question is
21:21 how do we map from a record to an auth record?
21:22 thd kados: bib records are mapped to auth records with $9 in each of the controlled fields
21:23 kados: However, you knew that so what had you meant by your question
21:23 ?
21:23 kados thats' not what I meant
21:23 I mean, how do we construct the auth record values from the bib recor dvalues?
21:24 thd kados: you can only construct 1XX in the authority records which is why that is the poor person's system
21:25 kados even if the record has values in 6XX and 7XX?
21:25 thd kados: the 1XX is in the authority record
21:25 kados: so starting from 650 ..
21:26 in a biblio you get 150 in an authority record.
21:28 kados: You cannot fill 4XX, 5XX in the authority records by building from bibliographic records.
21:28 kados: although you could try to extract the form of author names used on the material itself from the statement of responsibility
21:30 245 $c
21:31 kados: but that would help you only a little.  Authorities only do what they should when you import the NACO authorities.
21:32 kados: existing systems that use them well are rare to none
21:33 kados: most systems simply verify cataloguing against the authorised form after the cataloguing data has been entered.
21:34 kados we can do better than that :-)
21:35 thd kados: LC uses a system that checks existing uses and provides a drop down list after the cataloguer has already typed everything
21:35 well actually it is more like autocomplete
21:37 kados: but it does not help the cataloguer search against 4XX 5XX when typing
21:37 kados: Koha has that system beat already.
21:40 kados: or at least Koha would have that system beat if real authority records were in Koha
21:42 kados: I had a half reasonable understanding of how bulkmarcimport.pl worked a few months ago.
21:42 kados: we need to get build_authorities.pl working before we can marry the two.
21:43 kados in the display
21:46 thd yaay
21:47 kados anything to put off writing import scripts :-)
21:47 thd kados we need to fix Results for Search: 1009 = 3509
21:47 No results found.
21:47 kados ahh ...
21:47 which search does that happen on?
21:48 thd kados:Frontier and pioneer life Fiction
21:48 kados: There are multiple   Frontier and pioneer life Fiction when searching frontier
21:48 kados yep
21:49 poor algorythm for matching
21:49 well actually ... notice that they have subdivisions included
21:49 thd kados: where does the matching algorithm live?
21:49 kados build_authorities.pl
21:55 results for search fixed
21:56 :-)
21:57 thd kados : so what lines do the matching
21:57 ?
22:00 kados dunno I haven't really looked at it
22:01 thd kados: what happened to the various authority types from last night?
22:03 kados: I may see the problem for not finding the biblio records
22:04 kados: the hash is storing codes to identify the authority types
22:04 kados: These are in all capitals in French
22:05 kados: Those codes need to match the authority type.
22:05 maybe :)
22:07 I think that requires searching the UNIMARC framework code to find where they are used
22:11 kados thd: i deleted the other auth types
22:11 thd: because there should only be four search points
22:11 thd: I have to step out for a bit -- get some dinner
22:12 thd: when I get back I'll start working on an import script
22:12 thd: (the problem is that we dont' have any good auth records)
22:12 thd: (if you know of some we could use or want to download a bunch from LOC that might be helpful)
22:12 thd kados; and explain four search points instead of a few more
22:12 kados anyway ... I'll be back in a bit
22:12 thd: i will
22:14 thd kados: I have 17k authorities from LC
22:15 courtesy of pines
22:38 kados thd: back
22:38 I have those too
22:39 I also have the PINES data to go with them -- all 5 million records :-)
22:39 thd: have you looked at them yet?
22:40 ooh ... they're not bad
22:40 looks like lots of serials
22:42 LOC says you can download their MARC records for use in a library system
22:42 thd: I wonder if we can batch download them slowly over the course of many days to arrive a a complete set of authority records for a client :-)
22:44 thd kados: LC will log your IP address and block you :)
22:44 kados thd: no prob ... I've got about 400 IP addresses :-)
22:45 thd kados: you could use LWP to write a script
22:45 kados: I have done that but not for LC
22:45 kados: LC is a little tricky because of session initialisation
22:46 kados: I have seen that some people have overcome that issue
22:47 kados: I actually have some python script that I had used
22:48 kados: I had asked someone at CDS about doing that  sort of thing manually, knowing that I could script t
22:48 kados: The answer I was given was that there was a limitation on the number of records in a given period
22:49 kados: The only way that could really work though is by IP address.
22:50 kados: apart from contract access to some of their content
22:51 kados: yet it would still be incomplete because ...
22:52 there is no "Search access to form, genre, and topical subject subdivisions"
22:59 kados: there is a very large group of records that they still do not distribute through CDS even for money
23:01 kados: look at the beginning dates for the retrospective files from CDS
23:01 http://www.loc.gov/cds/mds.html#ba
23:05 kados: The older MARC records from retrospective conversion of pre-MARC printed cards have issues like racist subject headings that keep them out of CDS distribution until the updating is complete.
23:08 kados: ok the information needed to make such a script work is partly in the method UNIMARC uses for building authorities using a standard set of codes for authority types.
23:08 kados: why only 4 authority types?  You are missing some.
23:09 kados do we want more than 4 search points?
23:09 I've never seen an authorities search with all 7
23:09 thd kados: you can and should group them but there are more authority types in the standard
23:10 kados more than 7?
23:10 thd kados: More than 4
23:10 kados position 9 in 008 only has 7
23:10 and those 7 can be grouped easily into 4
23:10 they will still retain their types
23:10 because they all use different tags
23:10 thd kados: that is different usage of type that what I meant
23:11 kados what did you mean?
23:11 if we have more than four authority frameworks in Koha there will be more than 4 search points
23:11 which we don't want
23:11 thd kados: I was referring to authority frameworks linked to a 1XX $a in an authority record.
23:12 kados: you can build a search to search more than one.
23:13 kados: you can group the inde4xes together
23:13 kados what is the advantage to having so many frameworks in Koha?
23:14 instead of grouping the frameworks?
23:14 thd kados: It is not an advantage it is the existing design.
23:15 kados: Grouping would be better but there is no code for that yet.
23:15 kados that would be easier than changing the search I think
23:16 thd kados: There is already code for changing the search.
23:16 kados ?
23:16 where?
23:16 thd kados: The link function in the biblio frameworks.
23:17 kados I have yet to see that work
23:17 in fact I don't know what it does
23:17 thd kados: I tried to ask paul what it did precisely and although he was unclear in his memory he seemed to agree that my suggested likely use was correct.
23:18 kados also, that does not solve the problem that the search types are directly linked to the frameworks
23:18 when you go to do a search you have to select a type
23:19 thd kados: just as the see also column extends the string search the link column extends the value restriction search.  however that is not the same as an authorities search.
23:21 kados what is a 'value restriction search'?
23:21 thd kados: The same mechanism must be extendable to the authority frameworks, however, if a column is added to the bibliographic frameworks.
23:22 kados: value restricted is what the '...' search does in the OPAC now.
23:22 kados heh ... as if that does anything :-)
23:22 it's just like doing a search before you do a search
23:22 thd kados: It searches only the value chosen in the result
23:22 kados that's actually not true :-)
23:22 thd kados :0
23:23 kados that may be the way it _should_ work
23:23 but it certainly doesn't work that way
23:23 thd kados: only if you use the wrong query terms
23:23 kados if you can find a search where it works let me know
23:24 from what I've seen all it does is a normal search
23:25 thd kados: if you search for hot air as value restricted and choose hot air it is not designed well enough to give you only hot air and not hot air baboons.
23:26 kados: however, if you search for hot air as value restricted and choose hot air balloons it will not find merely hot air.
23:27 s/baboons/balloons/
23:27 kados: do you see that distinction?
23:28 kados in theory
23:28 thd kados: the search should be an exact field search except for problems with MARC in SQL.
23:28 kados but I've yet to see it working in Kokha
23:29 thd kados: That is because you always choose the hot air option when you are done and not the hot air balloons option
23:30 kados like i said ... if you can find an example ... I'd be very excited :-)
23:31 thd in which case the template should be fixed
23:32 kados looking at paul's current code
23:33 he seems to have intended that the authorities frameworks
23:33 have a one-to-one corolation to leader position 9
23:34 which from what I can tell is different than the headings
23:36 there are 7 possible leader values for position 9
23:36 however, there are 8 possible types of headings
23:37 thd kados I will check the UNIMARC use of the leader there which is different
23:38 kados: The link search works fine except for some template issue fro the right column JavaScript
23:39 s/fro/for/
23:39 kados can you show me an example?
23:40 thd kados: It is labelled authority search but it is not really since it works fine with no authorities built
23:40 kados: search for United States as a subject
23:41 kados: then choose any single result you get that result only
23:41 kados: initiate the search by clicking on '...' of course
23:42 kados how are the results any different than just typing in the values?
23:42 (also, bad example because there are only results with 1 record)
23:43 thd kados: the selection list has biblio counts but not biblios.
23:43 kados search on 'forest'
23:43 and select 'forest animals'
23:44 there should only be 4 reusults right?
23:44 but there aren't :-)
23:46 thd kados: I see I have six instead of 4 although it works for the part labelled authorities or seems to well enough
23:46 kados try it with the authorities values too
23:47 pick 'rain forest ecology'
23:47 not one result but three
23:47 thd kados: it may be in part that the linking is not correct
23:48 kados the problem is that the only thing the pop up dictionary search does is fill text values int he normal search box
23:49 thd kados: you have to use the values on the left column
23:49 kados ?
23:49 thd for the authorities search
23:49 kados I was picking 'select and close'
23:51 thd kados search forest and then choose the first authorised value Rain forest ecology -- Juvenile literature
23:51 kados: the JavaScript links on the right are not correct
23:52 kados: They can be fixed though
23:52 kados sure, that one works ... but it also works if you just type it in the text box
23:53 thd kados: but you did not have a list of matching subjects and only matching subjects from the textbox
23:53 kados: The textbox search returns a list of biblios.
23:54 kados: you would need to open each biblio to see the subjects
23:55 kados: the '...' search gives you a different results display that you can use to refine your search
00:00 kados: it does not seem to search tracings and references in the authority but that would only need the proper support of a framework column for that
00:01 kados: It is not a true authorities search
00:03 s/framework/biblio framework/
00:09 kados: leader position 000/09 is undefined in UNIMARC for both bibliographic and authorities.
00:09 kados: were you looking at the do nothing code in bulkauthimport.pl?
00:09 kados why does paul use it in bulkauthimport.pl?
00:10 why do you say it is do nothing?
00:10 it looks like it might work
00:10 all it does is call AUTHaddauthority
00:10 thd kados: That code was a couple hours start that was abandoned before he had anything right
00:11 kados: you cannot learn much that is correct from that code.
00:11 kados gotcha
00:12 thd kados: the code to study is build_authorities.pl and bulkmarcimport.pl
00:13 kados: If you understand how build_authorities.pl works then you can add authority supporting code to a copy of bulkmarcimport.pl to create a working bulkauthimport.pl.
00:14 kados both scripts use the same subroutine
00:14 AUTHaddauthority
00:14 thd kados: both which scripts?
00:15 kados build_authorities.pl and bulkauthimport.pl
00:26 thd kados: paul's vague intention with bulkauthimport.pl is obvious and that can be seen from the code.
00:28 kados: first you must test for whether you have an authority record or a bibliographic record.
00:28 kados: that can be determined from the leader.
00:29 kados I'm not sure that's the way to go about it
00:29 I think we already know which are auth records
00:29 thd kados: paul imagined that the leader also contained a clue about authority type but it does not.
00:30 kados: authority type is determined from the 1XX used.
00:30 kados right
00:31 thd kados: after that point his intention of labelling the type more explicitly in a single field may be useful.
00:31 s/field/local use field/
00:32 kados: beyond that bulkauthimport,pl has no utility for creating a working model.
00:41 kados so how do we break things down into our framework codes?
00:41 NAME, SUBJECT, NAME / TITLE HEADING, UNIFORM TITLE
00:44 X00  Personal names
00:44 X10 Corporate names
00:44 X11 Meeting names
00:44 X30 Uniform titles
00:44 X48 Chronological terms
00:44 X50 Topical terms
00:44 X51 Geographic names
00:44 X55 Genre/form terms
00:44 do those map?
00:46 thd kados: looks good to me
00:46 kados terms are subjects?
00:47 thd yes
00:47 kados how do we handle NAME/TITLE HEADINGS?
00:48 thd kados: those are in uniform titles
00:48 kados er?
00:49 what is the difference between NAME/TITLE HEADINGS
00:49 and UNIFORM TITLES?
00:49 thd kados: Shakespeare/Hamlet could be a uniform title
00:50 kados: then it would also serve as a name/title heading
00:51 kados: LibLime Newsletter could also be a uniform tile
00:54 kados: When it is transformed into the LibLime Journal of Information Science it may change 130 but then so would the older bibliographic entry for the record where 245 has LibLime newsletter
00:54 kados: then LibLime Newsletter would become a 530
00:57 kados: uniform title is also used for series such as Education through Art
00:58 so every volume of the Education through Art series has the same uniform title.
01:06 the series uniform title would be an 830 in the bibliographic record not 130 which would be for the individual volume if there were a uniform title for the individual volume.
01:12 kados what is the difference between NAME/TITLE HEADINGS and UNIFORM TITLES? :-)
01:13 are all TITLES and NAMES also in NAME/TITLE headings?
01:13 or only some of them?
01:13 or none of them?
01:15 under what circumstances on import do I create a NAME/TITLE heading instead of a NAME or a TITLE heading or should I always create all three?
01:15 thd: you still there?
01:15 thd yes
01:17 kados: the biblio will already have a name / title unless you are creating original records.
01:17 kados I thought we were talking about auth records ...
01:18 thd kados: the biblio will have the fields it has unless you are asking about how to catalogue a particular work
01:19 kados: I was talking about matching authority record content to biblio content by adding a $9 to a match in the biblio.
01:20 kados I don't understand why there exist name headings, title headings and name/title headings
01:20 why would you want a name/title heading instead of separate title and name headings?
01:20 thd kados: name/title headings are possible uniform title records.
01:21 kados: if the uniform title authority is in name /title form like Shakespeare/hamlet then it its a name title heading.
01:22 kados I don't think we can currently handle name/titles
01:22 we can only deal with them independently
01:22 if I'm not mistaken
01:22 thd kados: why not?
01:22 kados for several reasons
01:23 one, we can only map one thesaurus value to a given subfield
01:23 you can't map both NAME and NAME/TITLE too 100a
01:24 and you can't map both TITLE and NAME/TITLE to 245a
01:24 thd kados: The authority has the contents of the 130 match that or try to match to 130 in the biblo and any place else that might be relevant in case I overlook something.
01:25 kados: 245 is not controlled.
01:25 kados why not?
01:25 shouldn't a good system have full authority control where relevant?
01:25 thd kados: 245 has what is printed on the title page just as it appears.
01:26 kados: a good system would be full of 130s in bibliographic records which are rare in the real world.
01:27 kados: You wrote a fine book World Improvement by kados
01:27 and that is how it is printed on the title page.
01:28 kados: that book will have 245 $aWorld improvement$cby kados.
01:30 kados: the main author will be controlled 100 $aFerraro, joshua Mxxx$d19XX-
01:30 kados: some clever librarian wants to catalogue the French edition
01:31 kadso: the clever librarian creates a uniform title in 130 $aWorld improvement
01:33 kados: the French will have 245 $aImprovement monde or whatever that would be in French$cby( i should know the French for that) kados.
01:35 kados: the clever librarian could have made 130 $aFerraro, Joshua Mxxx / World improvement or however that would be done
01:35 kados but I thought a uniform title was only used when a title was not associated with a specific author, like the bible
01:36 thd then you would have a name/title authority as a uniform title
01:36 kados _as a uniform title_?
01:36 they are different things
01:37 a name/title authority can't _be_ a uniform title
01:37 can it?
01:37 thd kados: Uniform titles are for distinguishing the work or the expression across different editions
01:37 kados is there a difference between a uniform title and a uniform title heading?
01:37 thd kados: they are the same
01:38 kados: name/tiltle is a special kind of Uniform tilte
01:39 kados so if we're importing
01:39 thd kados: 130 $aShakespeare/Hamlet  would be different form 130 $aFerraro/Hamlet
01:39 kados is it safe to say that we can check to see if there is an author listed along with a title
01:39 and if there is, it is a NAME/TITLE
01:40 thd: I think the above example would be written:
01:40 100 $a Shakespheare,
01:40 thd kados: you check the value of 130 in the authority record against 130 and 830 in the biblio
01:40 kados $c hamlet
01:41 thd kados: yes they are not common and I had not checked
01:42 kados: uniform tile covers more than just 130 and 830 in the biblio also but I have not checked all the places
01:43 kados thd: for 'Bible' the LOC authorities only has 25 entries in NAME/TITLE
01:43 thd: none of them can be clicked on
01:43 there are also 25 subjct headings for Bible
01:44 same with title
01:44 same with NAME
01:44 thd kados: those are not the authority record if they cannot be clicked
01:44 kados that seems like too much of a coincidence
01:45 thd kados: names can also be subjects
01:45 kados i koha's current scheme, if you do a 'NAME search you will not get 'TITLE entries back
01:45 thd s/subjects/used as subjects/
01:45 kados same with a Subject Search
01:46 but it seems that with LOC no matter what search you do you get the same auth files back
01:46 thd kados: it is difficult to find the authorised forms because searches are left anchored
01:46 kados only it picks different parts of them out for display
01:47 thd kados: 600 is a subject based on a personal name
01:47 610 corporate name subject, etc.
01:49 kados: Bible could be in a 610 perhaps as well as a 650 it is too fundamental too know
01:50 kados: the links that you cannot click on in authorities.loc.gov can be clicked in catalog.loc.gov
01:50 kados thd: is this voyager?
01:50 thd: that stephen thinks is so great?
01:50 thd catalog.loc.gov also does not force left anchoring
01:51 catalog is voyager but authorities could be an in house or contract job.
01:53 kados: voyager like other OPACs is configurable or customisable.  catalog.loc.gov looked much the same before they ever outsourced it.
01:54 kados: before 2000 LC ran much of their own software.  Their catalog was mostly open source but it did not do Unicode.
01:55 kados it was open source?
01:55 thd kados: yes I will find the link
01:59 "the ISearch-CGI public domain software that is available from CNIDR. It should be noted that many search and retrieval capabilities that are available in the Z39.50 protocol are not implemented in this gateway. The Initialization, Search, and Retrieval facilities have been implemented."
02:01 http://www.cnidr.org/ is probably outdated but if you google for Isearch you can find the German company that maintained a very expensive commercial version which fixed the Unicode and other problems
02:03 CNIDR was an agency set up by the library of congress to create this for them.  They also were involved with the creation of WAIS if you remember that.
02:04 kados interesting
02:05 thd kados: I was going to use Isearch long ago until I discovered why LC abandoned it.
02:05 kados: It was designed for MARC records.
02:05 kados heh
02:08 thd kados: If Zebra did not exist Isearch would be work looking at but I have no C background.
02:09 kados: Zebra does much more than ISearch ever imagined.
02:10 kados: Also there were special LC modifications to ISearch so that it did what they wanted.  Those modifications were not part of the public domain distribution.
02:12 kados: prior to that most of their systems at LC were in house systems except cataloguing which is an MSDOS program made by someone whom I have spoken with who has an office in New York.
02:14 s/that/year 2000/
02:16 you can still see Isearch doing some of what it had done at http://www.loc.gov/z3950/
02:17 kados: you woke up too early this afternoon :)
02:20 kados: ISearch had a companion component ISite that was the backend and not part of the gateway where ISearch connects to other backends.
02:33 good night kados
04:23 I fixed author searching of authorities in the OPAC
04:24 I had started to do that a few hours ago and then distracted myself
04:25 kados: yay it works
04:25 kados: search authors '...' for jack lewis
04:46 kados: this was merely a bibliographic framework issue

← Previous day | Today | Next day → | Search | Index

koha1