about us >> what we do

Development of Expert Systems and Reasoning Engines

Because customer data is predominantly non-numeric it is not feasible to use the integrity and completeness checks that can be used for conventional data.  The “subjective” nature of customer data makes it very difficult to be understood by computers.

In the early 1980’s Kewan McDonald, the father of Harmony’s predecessors, determined that the problems with customer data were ideally suited to expert systems.  He reasoned that since the problems with customer data stemmed from the fact that they are written and read by people, the best way to solve these problems was to use a system that replicated the processes people use to interpret customer data.

One of Kewan’s contemporaries was Professor J. Ross Quinlan, who had established the university as a centre of excellence in expert systems and machine learning.  At the time Professor Quinlan was developing his “Interactive Dichotomizer 3” (ID3), one of the original decision tree based expert systems developed in the world.  ID3 and its successors C4.5, and C5, remain among the most widely used machine learning algorithms worldwide.

Kewan used early versions of the ID3 Decision Tree to develop a prototype matching system that replicated the processes people use to compare and assess names and addresses.  He developed MatchMaker, a decision tree processor using an external set of rules to perform the classification and comparison processes to consistently and reliably match customer data.

Kewan enjoyed the full support and mentorship of Professor Quinlan during the development of MatchMaker.  It was one of the first fully functioning expert systems at the time.

Mastersoft Research continues to develop products which employ the principles and foundations of expert systems and reasoning engines.  We have evolved the MatchMaker application into the Harmony Data Quality Series of products.

Learn more about the Harmony Data Quality Series of products

 

“With the introduction of Harmony, we have regained confidence in our data warehouse. The information is now of high quality, enabling more accurate targeting of our market segments. This will improve our service and expand our marketing capabilities.”