Best practice (Ex#26, 27):
These experiments were conducted to test the
combination of some of the best results from
the other experiments. Both Ex#26 and Ex#27
performed considerably better than the other
experiments with an increase of 13,1% and
10,6% respectively. The system performed
considerably worse for the Wine ontology in
Ex#27 than for Ex#26; this is probably due to
the ranking problem as described for Ex#22
and Ex#24. However, if we look at the error
rates (Table 6) for Ex#27, this is very low
indicating that additional relations provided by
a reasoner have a positive effect on populating
more entities (the same findings as for Ex#22).