Abstract
In the last years, online users have been sharing more and more opinions, reviews, and comments on the web. Opinion mining is the automatic process of getting the subject of such opinions, and recently it has been attracting great commercial and academic interest. Several methods were presented for performing opinion mining in Bangla language, however they reported limited performance. In the present article, we considered the only two publicly datasets available for opinion mining in the Bangla language. We machine translated the datasets into the English language and we preprocessed them by extracting textual frequency based features. Then, we designed two stacked contractive auto-encoders based architectures to perform opinion mining in Bangla language, one for each dataset. The classifiers were trained on the machine translated version on the two datasets in a stacked learning fashion. The proposed classifiers achieved improved performance, with respect to accuracy (ge 96%), precision (ge 93%), recall (ge 94%), and F1 score (ge 94%), reported in the past state of the art works. Furthermore, the experimental results showed that both the machine translation procedure and the stacked learning frameworks improved the final classification performance.
Highlights
Latest research works showed that Internet users sometimes trust more to online reviews than their relatives or friends (Lăzăroiu et al 2020)
The confusion matrices report the best performance reached in the case Vector Space Model (VSM) vectors were computed relying on Eq (7), and in the case when Contractive Auto-Encoders (CAEs) were trained with the stacked learning framework
The lowest performance were obtained for the “Bowling” class in the Cricket dataset and for the “Ambiance” class in the Restaurant dataset
Summary
Latest research works showed that Internet users sometimes trust more to online reviews than their relatives or friends (Lăzăroiu et al 2020). In order to prevent people from understanding the main drawbacks of their products and services, companies often want to limit online users’ participation in reviewing them. This behavior represents a crucial point of many big companies’ market strategy in a more and more competitive world (Trusov et al 2009). The most tackled task within the context of SA is the classification the polarity of SA can been carried at three main levels (Yue et al 2019; Li et al 2019; Hacid et al 2018): at the document level SA classifies the whole opinion expressed in the document with positive, negative or neutral sentiment. The task of ABSA is twofold: it first requires the classification of aspects on which the document is focused, and this sub-task is usually identified as opinion mining or
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: Journal of Ambient Intelligence and Humanized Computing
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
