Using a Semi-Supervised Machine Learning Approach for Sentiment Analysis and Topic Modeling
Open AccessDownloadable Content
Companies that seek to follow a customer-centric approach to product development and problem resolution often rely heavily on consumer feedback for discovery of product issues, recommendations for feature enhancements, and opinions on product benefits and functionality. Advances in data mining, natural language processing, and neural networks combined with an ever-increasing availability of consumer data enables companies to quickly sift through customer opinions, identify a number of opportunities and defects, and react in accordance with consumer needs. Using an automated method to gather consumer feedback, separate the feedback into positive and negative sentiment, and determine the product-relevant topics enables manufacturers an opportunity to react to consumer feedback in real-time. The methodologies and code developed for this Praxis were created to help manufacturers react to product issues and consumer needs by scraping Online Consumer Reviews (OCRs) from social media in real time, performing sentiment analysis, and identifying sentiment topics pertaining to a specific manufacturer’s product through the use of machine learning models and algorithms. Recurrent Neural Networks (RNN), Convolutional Neural Networks (CNN), and Bidirectional Encoder Representations from Transformers (BERT) models were used for Sentiment Analysis, while Topic Modeling was handled through Latent Dirichlet Allocation (LDA) and BERT models.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.