I'm forwarding this message which Joe Zeeman sent
to the Bath list (he didn't copy us. It's relevant
to the current discussion on truncation. See point
1.
--Ray

-------- Original Message --------
From: Johan Zeeman <[log in to unmask]>
Subject: Re: SRW cql Prefix for bath searches
To: [log in to unmask]

Some comments:

1. I would much rather have the query language
itself contain a truncation
syntax than have to define separate indexes for
each instance of a truncated
term. Something like "bath.author=Shakespeare,
W*" would be preferable to
"bath.authorRT=Shakespeare, W" because it
corresponds more closely to the
way I believe users construct queries.

2. I would prefer the keyword indexes (3 and 4)
to have "word" in them":
"bath.anyWord", bath.subjectWord"

3. I'm not sure I understand the logic behind the
different construction of
"bath.author.exact" and bath.authorRT". Either
you have "bath.authorExact"
or you have "bath.author.RT". I would probably
prefer "complete" to
"exact", since many bib systems have "exact"
indexes that do not in fact do
an exact match (they do a first words in field
match) and there is scope for
serious misundersanding and corresponding
inconsistent implementations.

4. I thought the author search with normalized
name was being withdrawn?

5. Would an index name "...FirstWordsRT" be
synonymous with
"...FirstCharacters"? The attribute combinations
certainly suggest so. If
so, I would prefer the former (assuming we retain
"RT" for right truncation)
as being parallel with the naming of the word
indexes.

j.


----- Original Message -----
From: "Ray Denenberg" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Tuesday, May 14, 2002 11:41 AM
Subject: SRW cql Prefix for bath searches


> The ZING SRW/U implementors (see
> http://www.loc.gov/z3950/agency/zing/zing.html)
> are defining a string query language, CQL (common
> query language), where an abstract index name
> (prefixed with the name of the index set, e.g.
> "bath", "dc") will be sent in place of an
> attribute combination.
>
> We are now defining these abstract indexes and
> we've drafted a set of abstract indexes
> corresponding to bath searches, at
> http://www.loc.gov/z3950/agency/zing/bath-prefixes.html
>
> For example in the cql string, "bath.author"
> would be sent in lieu of the attribute combination
> corresponding to a Bath author search.
>
> We would be happy to hear any comments that Bath
> implementors have on this approach.
>
> --Ray Denenberg