Comparative analysis of automatic keyword extraction algorithms: KeyBERT, YAKE and GPT-3

Polo Bautista, Luis Roberto and Casique Vasquez, Raquel Comparative analysis of automatic keyword extraction algorithms: KeyBERT, YAKE and GPT-3., 2024 [Preprint]

Preview

Text Comparative analysis of automatic keyword extraction algorithms KeyBERT YAKE and GPT 3.pdf Download (578kB) | Preview

English abstract

In this paper, a comparative analysis of three algorithms used for automatic keyword extraction was performed: KeyBERT, YAKE and GPT-3. The main objective is to analyze the cosine similarity between the automatically extracted keywords and those assigned manually as part of the indexing performed by the authors of the articles. To carry out this comparative analysis, a Google Colaboratory environment was developed and a corpus of English article abstracts obtained from Scopus was used. The results revealed that GPT-3 obtained a higher cosine similarity, demonstrating efficiency in the automatic extraction of keywords and generating results similar to how a person would do it.

Item type:	Preprint
Keywords:	Automatic keyword extraction, Cosine similarity, GPT-3.
Subjects:	I. Information treatment for information services > IB. Content analysis (A and I, class.) I. Information treatment for information services > ID. Knowledge representation. L. Information technology and library technology > LM. Automatic text retrieval.
Depositing user:	Luis Roberto Polo Bautista
Date deposited:	18 Feb 2025 19:11
Last modified:	18 Feb 2025 19:11
URI:	http://hdl.handle.net/10760/45969

Downloads

Actions (login required)

View Item