Onto.PT: Towards the Automatic Construction of a Lexical Ontology ...

7.1. Ontologising algorithms 115
Each algorithm can be seen as a different strategy for attaching terms a and
b, in {a R b}, to suitable synsets Ai ∈ T and Bj ∈ T , Ai = {ai0, ai1, ..., ain},
Bj = {bj0, bj1, ..., bjm}, where n = |Ai| and m = |Bj|. This results in a
sb-triple {Ai R Bj}. All algorithms, presented below, start by getting all
the candidate synsets from the thesaurus, which are those containing term a,
A ∈ T : ∀(Ai ∈ A) → a ∈ Ai, and all with term b, B ∈ T : ∀(Bj ∈ B) → b ∈ Bj.
Also, for all of the proposed algorithms, if T does not contain the term argument of a
tb-triple (e.g. a), a new synset containing only this term is created (e.g. Sa = {a}).
Before presenting the algorithms, we introduce figure 7.1, which contains candidate
synsets for attaching terms a and b, as well as a made up lexical network N.
There, nodes, identified by letters, can be seen as lexical items (terms), while the
connections represent tb-triples of a labelled type (R1, R2 and R3). Note that N
intentionally does not contain some lexical items in the synsets (k to p), which happens
if they do not occur in any tb-triple. Both the synsets and the network of
figure 7.1 will be used in the illustration of some of the algorithms. We intentionally
created an example where, depending on the used algorithm, the resulting sb-triple
is different.
Sample candidate synsets:
A1 = {a, c, d, k} B1 = {b, g, h}
A2 = {a, e, l} B2 = {b, f, o}
A3 = {a, m, n} B3 = {b, p}
A4 = {a, c, d, e, i, j}
Sample network N:
Figure 7.1: Candidate synsets and lexical network for the ontologising examples.
Related Proportion (RP): This algorithm is based on a similar assumption to
Pennacchiotti and Pantel (2006)’s anchor approach. First, to attach term a, term b
is fixed. For each synset Ai ∈ A, ni is the number of terms aik ∈ Ai such that the
triple {aik R b} holds. The related proportion rp is computed as follows:
rp(Ai, {a, R, b}) =
ni
1 + log2(|Ai|)
(7.1)