Abstract
Online social networks offer helpful information about people’s perspectives on a variety of topics. These data are gathered and evaluated by applications such as monitoring and recommendation systems (RS). The advanced wellbeing observation component of the knowledge-based recommendation system (KBRS) described in this work can help detect clients who might be stressed out or depressed. The KBRS, which is activated based on ontologies and sentiment analysis, sends signals of joy, relaxation, soothing, or motivation to people who are experiencing psychological challenges depending on the outcomes of the monitoring. A means for alerting appropriate parties is also part of the solution in case the monitoring system detects a depressed issue. The suggested strategy successfully distinguished between unhappy and focused clients with a precision of 0.89 and 0.90. Using a convolutional brain organization and bidirectional long short-term memory (BLSTM) – recurrent neural networks (RNN), depressing and upsetting sentences are detected. The recommended KBRS scored 94% of consumers who were highly delighted, according to trial results, compared to 69% for a RS that didn’t use sentiment metrics or ontologies. Furthermore, results from random tests revealed that the proposed method makes only modest demands on the processing, memory, and energy of modern mobile electronic devices.
doi: 10.17756/nwj.2023-s4-081
Citation: Sunitha L, Shreya K, Reddy BU, Yashwitha G. 2023. Enhanced Automatic Recommender System: Leveraging Sentiment Analysis and Deep Learning. NanoWorld J 9(S4): S483-S487.