Title: Automatic classification of tags in public opinion texts directed to companies in a Brazilian complaints web page

Authors: Marcony Montini de Oliveira Lima; Edwaldo Soares Rodrigues

Addresses: Department of Exact Sciences, Universidade do Estado de Minas Gerais, Divinópolis, Minas Gerais, Brazil ' Department of Exact Sciences, Universidade do Estado de Minas Gerais, Divinópolis, Minas Gerais, Brazil

Abstract: The expression of public opinion through digital means has become common since it helps to deal with problems regarding products/services directly with representatives of responsible companies through web pages such as Reclame Aqui in Brazil. There is a problem on this page generated by the large amount and redundancy of tags that characterise the complaints creating difficulty in separating classes of interest. The objective of this work was to carry out the automatic classification of complaint texts in tags, all in Portuguese, aiming to better organise the information, thus reducing redundancy. The definition of these companies and the parameters to be collected were essential in a previous analysis to determine the characteristics of the algorithms to be applied. Therefore, based on the analysis of the effectiveness of the processes and the adequacy of the dataset and the parameters, it was possible to fulfill the initial proposals of the work.

Keywords: classification; auto-tagging; Reclame Aqui.

DOI: 10.1504/IJISTA.2024.140960

International Journal of Intelligent Systems Technologies and Applications, 2024 Vol.22 No.3, pp.333 - 357

Received: 24 Sep 2023
Accepted: 15 Feb 2024
Published online: 04 Sep 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article