CORA is a probabilistic
relational data mining method. It discovers relational preference
patterns which determine when a complex object A precedes (in preference) another
object B and then it uses
these preference
patterns to estimate the probability of the preference relation for any
pair of complex objects. This probability is finally used to rank the
objects. The probability is computed by extending the naive Bayes
assumption to relational representations.
CORA is used to the domain of document image understanding for reading
order detection and for document summarization.
M.Ceci, A.
Appice, C. Loglisci, & Donato Malerba. Complex objects ranking: a
relational data mining approach. In S. Y. Shin, S.
Ossowski, M. Schumacher, M. J. Palakal, and C.C. Hung, editors, SAC,
pages 1071–1077. ACM, 2010.
M. Ceci, A. Appice, L. Macchia & D. Malerba. Relational Classification based on Emerging Patterns. Atti del XVI Convegno Nazionale su Sistemi Evoluti per Basi di Dati,
SEBD 2008. 21-25 Giugno, Mondello (Palermo) 2008. 45-56