Natural Language Processing

Word Sense Disambiguation 203

is called. The goal of the algorithm is to disambiguate each occurrence

of nouns and verbs in a text (corpus) T, and the descriptions of each

Procedure are:

• Procedure 1: Recognize the named entities in the text. From

now they are disambiguated words.

• Procedure 2: Identify the words in T which are monosemous

(one sense in WordNet) and disambiguate them with the single

sense.

• Procedure 3: If a target word (to be disambiguated) is in position

i in the text, denote it by W

i

. Form the pairs W

i–1

W

i

and W

i

W

i

+1

and extract from the tagged corpus SemCor all the tagged

correspondent pairs. If in these tagged occurrences the word

W

i

has the sense #k in a number which is larger than a given

threshold, then disambiguate ...