Dear Tibor, and dear all,
May I propose an old question (and possible problem) about that?
I see Tibor's explanation about colon character (": ") coming from
punctuation interposed between many subfields…
but It seems that (also in the beta version of Inspire) an old Invenio's
behaviour with space-colon-space (" : ") punctuation still remains.
Please take a look to the record
http://inspirebeta.net/record/852219/export/hm
titled (all in one 245__a string) "000852219 245__ $$aOrbital Parameters of
Binary Radio Pulsars : Revealing Their Structure, Formation, Evolution and
Dynamic History"
If you search (with "All of the words" option) for the title WITHOUT the
colon, you'll have the result (after some punctuation-cleaning):
http://inspirebeta.net/search?p=Orbital+Parameters+of+Binary+Radio+Pulsars++Revealing+Their+Structure%2C+Formation%2C+Evolution+and+Dynamic+History&f=&action_search=Cerca
if you search the title with the " : ", you'll not have the record:
http://inspirebeta.net/search?p=Orbital+Parameters+of+Binary+Radio+Pulsars+%3A+Revealing+Their+Structure%2C+Formation%2C+Evolution+and+Dynamic+History&f=&action_search
I already had an interesting explanation from Tibor (…some time ago:
06/11/2007…), referring that:
*CFG_BIBINDEX_CHARS_ALPHANUMERIC_SEPARATORS and
CFG_BIBINDEX_CHARS_PUNCTUATION variables only affect word indexes at the
time*
and also (quoting Tibor):
*The ':' character is special in that it is usually interpreted to stand for
logical fields:
title:cleopatra
If you mean that you would like to treat a search for
cleopatra:
just like a search for:
cleopatra,
then this requires some development: we can fix it e.g. in the way that
'foo:' will not be considered as a logical field candidate unless 'foo' is
defined in the logical field table. (The price to pay would be one SQL
lookup, but this can be cached.)*
Thanks for your attention,
Cristian Bacchi
On 3 May 2010 10:05, Tibor Simko <[email protected]> wrote:
> Hi Johnny:
>
> On Mon, 03 May 2010, Johnny Mariéthoz wrote:
>> when a title of a records contains one or more colon such as:
>>
>> http://cdsweb.cern.ch/record/1262161?ln=fr
>>
>> Title Recombination : An important effect in multigap resistive plate
chambers
>>
>> It is not possible to retrieve this record using the search engine of
>> invenio without removing the colon character.
>
> Actually, this record does not contain the colon character, the title
> and the subtitle values are stored in two different MARC subfields, see
> the MARC output:
>
> <http://cdsweb.cern.ch/record/1262161/export/hm>
>
>> I suspect that it is due that the colon character has a special
>> meaning for the search
>
> Yes, it has, but one can also search for it, e.g. on the INSPIRE site
> that is running the latest git/master sources, there is a record with
> the title:
>
> Properties of potential modelling three benchmarks: the cosmological
> constant, inflation and three generations.
>
> <http://inspirebeta.net/record/852915>
>
> You can search for it by entering for example:
>
> title:'benchmarks: the'
>
> <http://inspirebeta.net/search?p=title%3A%27benchmarks%3A+the%27>
>
> Best regards
> --
> Tibor Simko
>
