Applied Semantic Web Technologies

Data-Driven Evaluation of Ontologies  ◾  221

abstraction of the original input space. us, taxonomy guided learning algorithms

work on this induced input space.

9.2.2 Definition of Word Taxonomy (WT)

For word taxonomy over unstructured data such as text documents or sequences,

we dene abstraction based on the frequency of values associated with the same

class label.

Let Σ = {w

1

, w

2

, …, w

N

} be a dictionary of words, C = {c

1

, c

2

, …, c

M

} a nite set

of mutually disjoint class labels, and f

i,j

an integer frequency of word w

i

in a sequence

d

j

. Sequence d

j

is represented as an instance I

j

, a frequency vector < f

i,j

> of w

i

, and

each sequence belongs ...