Implement methods for one2many and many2one
assignment restrictions.
Implement more classifiers, for example the ORLS (Oxford
Record linkage Study) classifier, as well as improved machine
learning based classifiers.
Implement expectation-maximisation (EM) algorithm for parameter
estimation.
Include methods to calculate frequency counts from the data used
in a linkage process, not just external frequency tables.
Calculate positive predictive values as presented in the
paper Probabilistic record linkage and a method to
calculate the positive predictive value by Tony Blakely
and Clare Salmond, International Journal of
Epidemiology, 2002 31: 1246-1252. Available online at
http://ije.oupjournals.org/cgi/content/abstract/31/6/1246?etocPro
Febrl - Freely extensible biomedical record linkage
