Multinomial Naive Bayesian Classifier Framework for Systematic Analysis of Smart IoT Devices

Area of Science:

• Artificial Intelligence
• Machine Learning
• Natural Language Processing

Background:

• Businesses require accurate methods to gauge consumer satisfaction with products and services.
• Analyzing customer reviews is crucial for understanding market perception and improving offerings.
• Existing sentiment analysis tools may lack the precision needed for nuanced feedback interpretation.

Purpose of the Study:

• To develop and evaluate a deep learning model for automatic customer sentiment analysis.
• To accurately categorize user reviews of Amazon Alexa into positive or negative sentiments.
• To provide a scalable solution for companies to analyze online customer feedback.

Main Methods:

• A dataset of 3150 Amazon Alexa user reviews was collected and preprocessed.
• A deep learning model, specifically a multinomial naive Bayesian classifier, was implemented.
• The model was trained on 80% of the dataset and validated on the remaining 20%.

Main Results:

• The deep learning model achieved a high accuracy of 93% in sentiment classification.
• Initial analysis included plotting word clouds to visualize sentiment distribution.
• The proposed model outperformed three out of four benchmark models in the same domain.

Conclusions:

• The developed deep learning model offers a highly accurate and effective method for customer sentiment analysis.
• This approach can be readily adopted by any business with an online presence to automate review analysis.
• The research demonstrates the power of artificial intelligence and machine learning in understanding consumer feedback.