[221] in Information Retrieval
Review of Open Text Presentation.
daemon@ATHENA.MIT.EDU (Bill Cattey)
Thu Feb 3 14:43:06 1994
Date: Thu, 3 Feb 1994 14:42:05 -0500 (EST)
From: Bill Cattey <wdc@MIT.EDU>
To: elibdev@MIT.EDU, dlicc@MIT.EDU, nut-lib@MIT.EDU, alexp@MIT.EDU,
Summary: Open Text represents an attractive second search engine in
addition to Newton. It offers a database and clients of immedate use,
and which will soon fit into our larger architecture via Z39.50. Longer
term we would have two rather than a single viable database retrieval
server engine. I think we should buy it.
I attended Tuesday's presentation by David Terry of Open Text Co.
Here is my review of his presentation. Since this note is going to a
wide audience, I may repeat things that many of the readers already
know.
He told about the history, a little bit about the insides, and showed
examples of using the Open Text Software. This software consists of the
PAT Full Text Search Engine, the LECTOR Display system, and two search
clients which implemented different user interfaces to make Queries to
PAT.
The Oxford English Dictionary is one of the most prominent databases
served up by Open Text.
Considerable effort has gone to making the PAT search engine friendly to
even the most baroque document formats. Inside, PAT uses SGML which is
an open standard, so after you load a database into PAT, if you change
database engines, you're not stuck with files in some proprietary format
you can't decypher. Full text databases like Encyclopedias are becoming
available in ASCII and SGML.
PAT looks like an interesting engine with which to deliver these databases.
Why is it important to consider buying PAT when we already have Newton
and BRS/Search?
To review, Newton was acquired because it offered the ability to serve
data via the Z39.50 protocol instead of BRS's proprietary API, and most
importantly, we could license Newton for the whole campus for less than
it woudl cost to license BRS/Search on a SECOND host. Newton was
clearly a good thing to buy, and we have had to radically cut back our
expectations for BRS/Search.
The price of Open Text is about the same as Newton.
Open Text offers the following things that Newton does not:
An easier process to design and load databases.
Microsoft Windows based clients.
(Plans are to offer Macintosh Clients but timeframe is unclear.)
Preloaded Oxford English Dictionary
Converter programs to load a PLETHORA of existing
word processing formats.
Character based rather than key word based searching.
A product focused on Data users, not Programmers.
One glaring defect in Open Text is that it will not support the Z39.50
protocol until June.
I have worked closely with OCLC on Newton for the past year. It is a
good engine, but it is very hard to design databases for it, convert
input files to feed it, and to load the converted files. It represents
a non-trivial engineering development task every time we want to bring
up a new database.
Open Text represents to me a product that would allow people other than
Bill Cattey and Tom Owens to feed databases to.
The Oxford English Dictionary Database and the Motif and Windows clients
represent pieces that, if deployed today would be immediately useful to
the MIT community. There is the promise that four months in the future
the PAT database would interoperate with Z39.50, and could be made to
hook into a larger campus-wide architecture.
What is the cost?
The Oxford English Dictionary is $10K for a 5 year license, and 2.5K per
year after that.
Open Text is currently priced at $18K and 3.2K per year maintenance.
David Terry expressed interest in cutting a couple thousand off the
purchase price, or extending the warantee period beyond 3 months if we
act soon.
Open Text PAT currently runs on Decstation Ultrix, AIX, Solaris 2 and
SunOS in addition to HP-UX, and whatever SGI runs on their boxes.
Our cost on Newton was roughly the same, but we have still only received
the RS/6000 version. The other ports were promised but not yet
delivered.
It may be an act of simple prudence to get Open Text and hedge our
Database Server Bets.
-wdc
