The process of identifying records in two or more
data sets that share similar characteristics.
Information technology has allowed organisations to scrutinise details from very large data stores. The process of checking one record off against another is naturally quite laborous, so data matching typically would use special software which indexes data to ease in retrieval, or checks the efficiency of join conditions a programmer might make in SQL. Previously COBOL was used for data matching, although organisations now prefer to use relational database management systems such as DB2 and Teradata, as it makes it easier to compare related sets of data.
Data matching is used for many different purposes - government revenue protection, marketing, counter-terrorism, law enforcement, inventory control etc. You may have seen data matching occuring in films like Enemy of the State, except that the data matching computers where I work don't make tinny noises when you enter queries, or blippity noises when the results are displayed on the screen, or even barp! barp! noises with a blinking message whenever Osama Bin Laden's Amex card is swiped at a 7-11.
As organisations now desire to process data in real time in order to be more reactive to customer demands, matching needs to be made at faster speeds, with less down time. There is now also a greater public awareness of the privacy implications of data matching, since by combining two or more pieces of data together more can be learnt about a particular person. The licence plate of the idiot who cut off a tax auditor heading to work could be datamatched with the registered vehicle owner, and then with his name it could be datamatched with his tax records.
Data matching is not data mining, as the patterns the program will look out for will typically be static, and that more than one data set is required.