Semantic network

Definition

Semantic network indices capture lexical relationships such as polysemy (multiple meanings per word) and hypernymy (hierarchical relationships between general and specific terms, e.g., flower vs. rose).

Methodology

TAALES calculates semantic network indices using path-based measures in WordNet. Scores reflect either the number of senses a word has (polysemy) or the distance between concepts in a semantic hierarchy (hypernymy). Indices are averaged across parts of speech and path combinations.

Corpus Used

  • WordNet (Fellbaum, 1998)

Calculated indices

Polysemy

  • Indices:
    • content_poly
    • poly_noun
    • poly_verb
    • poly_adj
    • poly_adv

Hypernymy – Nouns

  • Indices:
    • hyper_noun_S1_P1
    • hyper_noun_Sav_P1
    • hyper_noun_Sav_Pav

Hypernymy – Verbs

  • Indices:
    • hyper_verb_S1_P1
    • hyper_verb_Sav_P1
    • hyper_verb_Sav_Pav

Hypernymy – Combined (Nouns & Verbs)

  • Indices:
    • hyper_verb_noun_s1_p1
    • hyper_verb_noun_Sav_P1
    • hyper_verb_noun_Sav_Pav

Notes

  • S1: Uses the first (most frequent) sense of the word.
  • Sav: Averages across all senses of the word.
  • P1: Uses the first (most direct) path in the hypernym hierarchy.
  • Pav: Averages across all possible paths for the word.

References:

  • Fellbaum, C. (Ed.). (1998). WordNet: An electronic lexical database. MIT press.