
156 Big Data Computing
Enabling access by object. The data structure designed for loading HDT-
encoded data sets, considering a subject-based order, is not suitable for doing
accesses by object. All the occurrence of an object are scattered throughout
the sequence
and we are not able to locate them unless we do sequential
scan. Furthermore, in this case a structure like the Wavelet Tree becomes
inefcient; RDF data sets usually have few predicates, but they contain many
different objects and logarithmic costs result in very expensive operation.
We enhance HDT-FoQ with an additional index (called O-Index), that is
responsible for