Analyzing Sentiment of Movie Reviews in Bangla by Applying Machine Learning Techniques

Document identifier: oai:DiVA.org:ltu-75891
Access full text here:10.1109/ICBSLP47725.2019.201483
Keyword: Natural Sciences, Computer and Information Sciences, Media and Communication Technology, Naturvetenskap, Data- och informationsvetenskap, Medieteknik, Bangla sentiment analysis, Support Vector Machines, Long Short Term Memory, Pervasive Mobile Computing, Distribuerade datorsystem
Publication year: 2019
Relevant Sustainable Development Goals (SDGs):
SDG 16 Peace, justice and strong institutions
The SDG label(s) above have been assigned by OSDG.ai

Abstract:

This paper proposes a process of sentiment analysis of movie reviews written in Bangla language. This process can automate the analysis of audience’s reaction towards a specific movie or TV show. With more and more people expressing their opinions openly in the social networking sites, analyzing the sentiment of comments made about a specific movie can indicate how well the movie is being accepted by the general public. The dataset used in this experiment was collected and labeled manually from publicly available comments and posts from social media websites. Using Support Vector Machine algorithm, this model achieves 88.90% accuracy on the test set and by using Long Short Term Memory network [1] the model manages to achieve 82.42% accuracy. Furthermore, a comparison with some other machine learning approaches is presented in this paper.

Authors

Rumman Rashid Chowdhury

Department of Computer Science and Engineering, University of Chittagong, Chittagong, Bangladesh
Other publications >>

Mohammad Shahadat Hossain

Department of Computer Science and Engineering, University of Chittagong, Chittagong, Bangladesh
Other publications >>

Sazzad Hossain

Department of Computer Science and Engineering, University of Liberal Arts Bangladesh, Dhaka, Bangladesh
Other publications >>

Karl Andersson

Luleå tekniska universitet; Datavetenskap
Other publications >>

Documents attached


Click on thumbnail to read

Record metadata

Click to view metadata