Pitfall of Google Tri-Grams Word Similarity Measure

Authors

  • Linda Wong Lin Juan Faculty Computer Science and Information Technology, Universiti Malaysia Sarawak, Sarawak, Malaysia.
  • Bong Chih How Faculty Computer Science and Information Technology, Universiti Malaysia Sarawak, Sarawak, Malaysia.
  • Johari Abdullah Faculty Computer Science and Information Technology, Universiti Malaysia Sarawak, Sarawak, Malaysia.
  • Lee Nung Kiong Faculty Computer Science and Information Technology, Universiti Malaysia Sarawak, Sarawak, Malaysia.

Keywords:

Google Tri-grams, Pitfalls, Sentence Similarity, Text Similarity, Trigrams, Unsupervised, Word Similarity,

Abstract

This paper describes and examines Google Trigram word similarity based on Google n-gram dataset. Google Tri-grams Measure (GTM) is an unsupervised similarity measurement technique. The paper investigates GTM’s word similarity measure which is the state-of-the art of the measure and we eventually reveal its pitfall. We test the word similarity with MC-30 word pair dataset and compare the result against the other word similarity measures. After evaluation, GTM word similarity measures is found significantly fall behind other word similarity measure. The pitfall of GTM word similarity is detailed and proved with evidences.

Downloads

Published

2017-12-07

How to Cite

Juan, L. W. L., How, B. C., Abdullah, J., & Kiong, L. N. (2017). Pitfall of Google Tri-Grams Word Similarity Measure. Journal of Telecommunication, Electronic and Computer Engineering (JTEC), 9(3-11), 43–46. Retrieved from https://jtec.utem.edu.my/jtec/article/view/3180

Most read articles by the same author(s)