A Study on the use of Etymology for Semantic Knowledge Extraction

Abstract

Etymology is the study of the composition of words through their historical roots. It is a rich area of study that dates back millennia, and that has contributed significantly to our understanding of human cultures and languages. The field of computational linguistics is a much younger field that grew from the advent of the digital era

and that has advanced continuously, even nowadays with the changes brought by Artificial Intelligence and Machine Learning. Computational linguistics have not yet leveraged the knowledge of etymology to its full potential. This work is a step to make etymology another contributor to the field of computational linguistics.

In this work we propose a framework to capture the complex etymological relationships that exist in the vocabulary of a human language by creating a complex network that associates words with their historical roots. We then use this framework to obtain insights into the semantics of the words that are part of the Chinese and Korean languages. We run two tasks: one of supervised learning, and one of unsupervised learning, and show that etymology can be effectively used to extract knowledge.

We believe that this work helps push etymology into the main stage of computational linguistics, and natural language processing.