SQLMatch - The accuracy of probabilistic linkage with the power of SQL

Linkage with no limits:

Most linkage programs on the market today force you to use proprietary macros, and work
with your data according to their rules.  Now, you can link your data using the popular
set-based language SQL, and Microsoft's very flexible relational database, SQL Server.

The Advantages of SQLMatch
SQLMatch is a set of Microsoft SQL server stored procedures and associated
tables in a linkage database.  It uses database industry standard SQL code, so any
SQL programmer will be able to perform true probabilistic record linkage without
embarking on a programming project.  AutoMatch and other popular linkage
programs use specialized languages and data structures.

SQLMatch does not require a copy of SAS or other statistical program.  So  
workstation and employee time that is usually devoted to statistical analysis is not
tied up with linkage tasks that are better suited for database programmers and true
relational database systems.

Each linkage project can be custom designed according to your specifications
before you receive the programs and database.  This is free of charge the first time
around.  You provide about 20 records from the source and target tables and you
will receive a SQLMatch database ready for you to import your full-sized tables using
the VB.NET front end Control Center program.  It is also easy to set things up with
no outside help if you wish.

Since it runs on MS SQL Server, data can be manipulated without restriction and
can be integrated into your organization in a standard sort of way.  Data views
accessible from within SAS, Excel or many other programs can be easily
accomplished, and data can be imported and exported in any way.  So in effect, you
also get the use of the fully relational database, MS SQL server, which is not
possible with other linkage systems including ones that are SAS based.

SQLMatch is a SQL stored procedure source code generator, which can create
entirely new and fully documented linkage programs.  The code is completely
customizable by any SQL programmer, and is only limited by the SQL language
itself.  AutoMatch and others only allow you to use the screens that they provide and
are limited to a small set of macro commands.  So they are like black boxes.  They
have no ability to generate new custom programs, though you can have several
linkage projects and go back and forth between them.

Hand review and QC is easier because the results are placed into a SQL server
table structured in such a way as to be easily browsed and viewed either from
within SQL server or in the MS VB.NET Control Center program that is included.  It is
easy to 'flag' the matches as 'good matches' and eliminate the bad matches.  And it
is easy to disseminate this information to any user on your network without
exporting, importing or converting any data.  Just provide a view.

It also comes with functions that you can use to clean your address data.  They
parse out addresses into a number, street name, suite/apartment number, and a
PO box number.  

Another function is a Jaro-Winkler string distance comparator.  With it, you can see
how closely two non-exact strings match.  This function can be used alone and is
also implemented within the program.  After the regular probabilistic linkage, fuzzy
match weights are calculated in conjunction with their relative rarity in your
database.  So fuzzy weights are also probabilistic fuzzy weights.  A close match on a
transposition of a common name like Maria and Marai does not score as high as a
rare one like Evelyn and Evelny.

Price.  Some probabilistic linkage programs sell for well over $100,000 with tens of
thousands of dollars in 'annual maintenance fees' per year per workstation.  
Consulting firms can charge hundreds of thousands or millions of dollars for
merely a moderately complex linkage. And the free programs use SAS, which is not
a relational database and most database programmers are unfamiliar with it.  So
there are hidden costs with these free programs.  The customized SQLMatch
programs, can be created in a few hours, and cost a few thousand dollars per
workstation for unlimited linkages with unlimited records.   The initial setup is done
for you if you like, and you can just import your sensitive data and run the program
on your local secure MS SQL server.

If you want to do more than generate and run the stored procedures, complete
source code is available for both the SQLMatch stored procedure generator (a set of
encrypted stored procedures in the standard version) and the VB.NET Control
Center program.  With full source code, you can integrate all of your existing code
with SQLMatch.  You may have proprietary procedures for importing or cleaning data
that you wish to seamlessly include in the project.  But source code is often not
necessary because you can use any stored procedure within the SQLMatch
database, since it is just a standard SQL Server database.
SQLMatch(tm) Probabilistic Linkage Program