Domain identification (Ex#16-21):
Not surprisingly, we got more or less the same
results as for the query expansion experiments
(Ex#2-Ex#8) where using parents, children,
and other neighbouring entities provided the
best results (Ex#21). Ex#19 got the same
results as Ex#21 but with a higher standard
deviation indicating that Ex#21 provides
better results independent of the ontology
quality. Ex#18 and Ex#19 provide more or
less the same results. For Ex#16, Ex#17, and
Ex#20 the algorithm failed to populate most of
the entities (see Table 6 and Figure 3). In fact,
for Ex#17 no entities were populated for the
Animals ontology since the ontology only got
super- and sub-class relationships and hence
no other relations. Consequently, the results
from Ex#16, Ex#17, and Ex#20 can be
disregarded.