Aplicaciones de inteligencia artificial para la clasificación automatizada de propósitos comunicativos en informes de ingeniería

Translated title of the contribution: Applications of artificial intelligence to automatic classification of communicative purposes in engineering reports

Research output: Contribution to journalArticlepeer-review

1 Scopus citations

Abstract

The task of recognizing the discursive patterns that enable the members of an academic discursive community to achieve the communicative purposes of the academic disciplinary genres is relevant for the processes of incorporation of new members through academic literacy processes. Research on academic genres has focused on expert genres and, to a lesser extent, on genres produced by university students, especially in engineering. The analysis of discursive genre patterns and artificial intelligence (AI), through Natural Language Processing (NLP) have had a complementary development, thanks to both automated classification algorithms and the increased availability of large amounts of textual data, which has enabled the classification of academic discursive genres. The aim of this paper is to automate classification of macromoves (MM) of the meso-genre report of practical experience in engineering. For this purpose, seven traditional classification algorithms, the deep learning model for Spanish called BETO and their corresponding configurations were considered. Among the findings, the best overall performance of SVM_lineal stands out. Also, SVM_linear, BETO and KNN are more effective for the classification of moves in different MMs. These results suggest that the combination of algorithms would be a useful procedure to better classify the macro-purposes of this meso-genre. It is planned to evaluate these algorithms in a tool for genre-based feedback of written production.

Translated title of the contributionApplications of artificial intelligence to automatic classification of communicative purposes in engineering reports
Original languageSpanish
Pages (from-to)942-970
Number of pages29
JournalRevista Signos
Volume54
Issue number107
DOIs
StatePublished - 2021
Externally publishedYes

Fingerprint

Dive into the research topics of 'Applications of artificial intelligence to automatic classification of communicative purposes in engineering reports'. Together they form a unique fingerprint.

Cite this