Fusing Sentiment Analysis with Hybrid Collaborative Algorithms for Enhanced Recommender Systems
Anindya Nag, Md. Mehedi Hassan, Mohammad Abu Tareq Rony, Biva Das, Riya Sil, Prianka Saha, Pronab Sarker, Anupam Kumar Bairagi · 2024
Recommender systems must be enhanced as Internet data grows to produce a list of suitable favorites. Given the exponential growth of Internet data, it is imperative to develop and enhance recommender systems to curate a personalized list of suitable preferences. The increasing prevalence of social networking sites has contributed significantly to the surge in popularity of review-based recommender systems in recent times. The rationale underlying such systems is to effectively utilize the knowledge acquired from written feedback provided by users. The fields of sentiment analysis and opinion mining have seen substantial growth in recent years. Consequently, there is now more work being put into improving recommendation systems. A hybrid collaborative filtering (HCF) recommender system will be used in this inquiry to provide a comparative analysis of sentiment analysis. The methodology being examined comprises two distinct phases, namely the sentiment analysis phase and the recommendation phase. The initial stage, referred to as “sentiment analysis,” entails the computation of sentiment scores through the utilization of datasets sourced from diverse social media platforms. The subsequent stage entails the utilization of HCF. The findings of this study demonstrate a noteworthy enhancement in the efficacy of recommender systems through the integration of sentiment analysis and other comparable systems. In addition, a comparison of the effectiveness of HCF methodologies and conventional methods often used in sentiment analysis is being conducted. The results show that, with a recorded accuracy rate of 99.2%, the HCF recommendation algorithm for sentiment analysis put forward in this study outperforms current state-of-the-art approaches.