Restaurant Recommendations with Implicit User Behavior via Image Classification and Sentiment Analysis from Social Media

Traditional recommendation systems often fall short in understanding users' true food preferences on social media platforms. This paper introduces an innovative approach to enhance the dining experience by leveraging implicit user behavior via image classification and sentiment analysis. The proposed system combines food image classification (MobileNetV2) and sentiment analysis (Transformer Flair) on social media comments, creating a robust framework for implicit recommendation. The results demonstrate the system's efficacy in accurately predicting food categories and extracting positive sentiments from user comments. By systematically linking these insights to user profiles, the paper establishes a recommendation system that considers both culinary preferences and geographical location. The seamless integration of computer vision and natural language processing allows for a more tailored and personalized restaurant suggestion, addressing the limitations of existing recommendation systems. The research highlights the potential of harnessing implicit user data from social media to revolutionize the landscape of personalized dining recommendations, setting the stage for future advancements in understanding and adapting to users' evolving preferences.