Resumo

BRAHIMI, Belgacem. Arabic Text Mining for Price Prediction of Used Cars and Equipment. Comp. y Sist. [online]. 2022, vol.26, n.2, pp.1015-1025.  Epub 10-Mar-2023. ISSN 2007-9737.  https://doi.org/10.13053/cys-26-2-3185.

Nowadays, companies and businesspersons are increasingly interested in the web for its potential and opportunities in marketing and commercial activities. Despite the importance of Internet advertising of used goods available on the web, work targeting their analysis is still limited. It is crucial for both buyers and sellers to precisely estimate the price of used products available online. Textual information that describes second hand goods is very relevant for accurate price prediction, however common solutions use only structural features for price estimation. We study in this paper the utility of using text mining techniques as well as the role of textual data integration in improving price prediction of online classifieds in Arabic. In order to evaluate the proposed methods, we collected online advertisements for two cases: used cars and lots of construction equipment. Additionally, we applied prediction algorithms to estimate the prices, namely, regression-based algorithms, K– nearest neighbor and neural network. Experimental results showed that the integration of textual features in the prediction models improves significantly price prediction compared with baseline methods that use only structured features. The results proved also that regression models are the best option for price estimation.

Palavras-chave : Text mining; supervised machine learning; regression; used cars prices; used equipment price; price prediction.

        · texto em Inglês     · Inglês ( pdf )