Semantic Similarity based on Weighted Ontology - Geographical Information Systems: Trends and Technologies - page 308

According to the variability approach, the weights are determined as
follows:

p v att
p v att + p v part

w att =

p v part
p v att + p v part
Whereas, according to the commonality approach, they are:

w part =

p c att
p c att + p c part

w att =

p c part
p c att + p c part

w part =

Example 7: In our running example, the number of geographic classes is
n = 7, in the case of parts m = 7, and in the case of attributes m = 13. Note that
the number of occurrences of attributes and parts are 25, and 9, respectively.
Thus, we have the following weights:

w att =0.470, w part =0.530
w att =0.599, w part =0.401
Let us consider Municipality, and County shown in Table 1. We assume
that S and T are the sets of attributes of Municipality and County , respectively.
Since, similar to Dice's function, the pairs of synonym attributes between
them are ( population , inhabitant ), ( area , surface ), we have:

|S ∩ T| =2

And the set difference are:

S-T ={ identity, councilHead,countryCode }
T-S ={ country ID }

Thus,

ܵ ௔௧௧ ሺܯݑ݊݅ܿ݅݌݈ܽ݅ݐݕǡܥ݋ݑ݊ݐݕሻൌ ȁ ܵځܶ ȁ

| S ⋂ T | + 0.4| S - T | + 0.6 | T-S |

ȁሺܵځܶሻȁ൅ͲǤͶȁሺܵെܶሻȁ൅ͲǤ͸ȁሺܶെܵሻȁ

| S ⋂ T | + 0.4| S - T | + 0.6 | T-S |

ൌͲǤͷʹ͸

For parts we have:
S part ( Municipality, County ) = 0