Aspect Based Call Analysis of Marathi Regional Language in India Using Machine Learning Approach

This paper discusses the importance of sentiment analysis in call centers, particularly in Marathi, a regional language in India. In this paper, Call center aspect based analysis system highlighting the importance of sentiment analysis in call centers, the challenges posed by Marathi language calls sentiment analysis as well as displaying a call-by-call summary of each call in Marathi. The Latent Semantic Analysis Algorithm (LSA) is used in the text summarizing approach and mathematical way that underpins it is Singular Value Decomposition (SVD). The Naive Bayes classifier and SVM are used to categories complete texts during semantic analysis by assigning them a label from a list of labels. The experimental results shows that the algorithm attributes Na¨ıve Bayes gives most appropriate result than SVM algorithm. The experimental results show that Na¨ıve Bayes algorithm provides the most appropriate results than SVM algorithm. Large call centers handle thousands of phone calls daily and struggle to maintain compliance with customer-agent interaction guidelines and legislation. Aspect-based sen-timent analysis (ABSA) provides a finer-grained understanding of sentiment by identifying the precise elements or traits being discussed. This is particularly useful in determining emotions based on public expressions. The paper highlights the challenges faced by call centers in analyzing Marathi language calls and analyze the product aspect based on customers sentiments.