Naslov
A Comparison of Machine Learning Algorithms in Opinion Polarity Classification of Customer Reviews

Autori
Žubrinić, Krunoslav ; Miličević, Mario ; Sjekavica, Tomo ; Obradović, Ines

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
International Journal of Computers, vol.3, 2018 / Bojkovic, Z. - : WSEAS Press ; IARAS, 2018, 159-163

Skup
18th International Conference on Applied Computer Science (ACS '18)

Mjesto i datum
Dubrovnik, Hrvatska, 26.09.2018. - 28.09.2018

Vrsta sudjelovanja
Pozvano predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
opinion polarity classification ; sentiment analysis ; natural language processing

Sažetak
In this paper we analyze reviews written by customers of an online shop, by employing opinion polarity classification on document level using five machine learning algorithms: Na¨ive Bayes, Support Vector Machine, Neural networks, C4.5 algorithm and classifier based on maximum entropy. We achieved the best results using Support Vector Machine algorithm (accuracy=0.845) and maximum entropy classifier (accuracy=0.84). Although those results are not as good as results that can be achieved in topic-based categorization, compared to similar researches in opinion polarity classification, they indicate a relatively good predictive performance of classical machine learning algorithms.

Izvorni jezik
Engleski

Znanstvena područja
Računarstvo

POVEZANOST RADA