Improving the soundex algorithm in conjunction with Paxus Health Ltd.

Searching a database of people's names is prone to problems from many sources. Some problems are caused by operator errors on the part of the person querying the database, such as using an incorrect spelling for a name that is given over the telephone (e.g. "McKay" instead of "MacKay"). Another problem is data entry errors at the time of entry into the computer, such as accidentally transposed characters. These problems decrease the effectiveness of searching the data, and so techniques need to be introduced to give reasonable results to queries even in the presence of multiple differences between data and search key. This project outlines some of the points that need to be considered when choosing an algorithm to perform approximate matching. Some of the methods available are described and evaluated, and the results are presented