Automatic keyword extraction from text with standard computer procedures

Slobodan Beliga

Abstract


Automatic keyword extraction takes a great interest as a research issue in the field of natural language processing and information retrieval. Although numerous methods for keyword extraction task have been developed, their effectiveness depends on many factors such as the approach used in method development, the domain to which they are adapted, the type of language or tasks for which they are constructed, etc., and still, there is a room for progress and improvements. In this paper, two existing methods are explained and reconstructed - RAKE and MAUI, which are the standard representatives of the unsupervised and supervised group of keyword extraction methods. It was experimentally tested whether the methods could successfully extract keywords from texts written in Italian, which had not been tested so far. For the experimental purposes, Italian texts were collected and annotated with keywords. The effectiveness of the MAUI method proved to be more promising than the RAKE method, which was confirmed earlier in the keyword extraction experiment from texts written in English.

Keywords


automatic keyword extraction; standard methods; RAKE; MAUI; Italian language

Full Text:

PDF (Hrvatski)


DOI: https://doi.org/10.36978/cte.4.1.1