The Marvellous Mechanical Matching Engine --- Help Page

This page describes how to run the Mmme software. For descriptions of the algorithms it uses consult papers on the subject available at http://www.richardclegg.org/pubs/.

Data format expected

The data is expected to be in plain text files with each line of the file being an observation. Only the first "string" on a line is taken to be an observation with the rest considered as comments on the data. Lines beginning with a # are ignored as comments. An example file is shown below:

# Data gathered on HIGH STREET 12/4/03 12:43 -- 14:43
ABC123X 12:43  TAXI
DGA543Z 12:44  HGV
OAS12Y  12:44
.
.
.

File Menu

This is for loading and saving of configuration data. The data saved is the error rates and probabilities for matching. The commands, Open, Save and Save As perform the expected actions on configuration files which are plain text files usually ending in .mme containing data about the error rates and matching probabilities.

Data Menu

This allows users to add new files of observations. Add allows the user to choose a new file to make a match with --- the file must have the format discussed above. Remove, removes a file from a potential match and Purge removes all such files. Individual files can be checked and unchecked in this menu to get a selective match between just a few files. If two or more files have been read in the the Match selection does the work of performing a match. Raw matches are output followed by matches corrected for false matches (this should be lower than the raw matches) and matches corrected for errors in recording (this should be higher than the previous corrected figure).

Options Menu

This menu allows you to set the error rates and probabilities of false matches. The probabilities menu allows you to set p(2), p(3) up to p(8). These probabilities can be set individually or to default values. The York default values are callibrated for data sets observed in the city of York (UK) in 2000 and 2001. The exponential fall off button allows you to set the probabilities to fall-off by a factor of p(2) from the last edited probability. The default setting is to have p(2)= 0.0001, p(3) the square of this, p(4) the cube and so on.

The Error rate menu allows you to set the individual rate of erroneous recording at each site as a fraction.

For more information about this project go to http://www.richardclegg.org/matching/