Exploria: A Deterministic, Sentiment-Aware Hybrid Restaurant Recommender for Low-Data Campus Environments

Recommendation systems are vital for assisting users in choosing the right services; however, the traditional methods use only numeric ratings that do not always capture user sentiment accurately, particularly in low-data, campus-centric environments. In this paper, a new restaurant recommendation system is presented, called Exploria, and it is a full-stack, deterministic sentiment-aware hybrid system designed to help college students in semi-urban areas. Exploria works by combining Content-Based Filtering (CBF), Item-Based Collaborative Filtering (IBCF), and VADER-based sentiment analysis, along with Bayesian smoothing used to adjust the sentiment scores in such a way that they become less sensitive to variations and more stable ranking even when the number of reviews is very low. Exploria is a hybrid recommender that has been tested on a low-data campus dataset that was manually curated and also included synthetic users, thereby serving as a bridge between algorithmic design and full-stack implementation, while enabling reproducible evaluation without the need for large-scale user interaction logs. In the low-data campus dataset, the conventional quantitative evaluation metrics of measuring the effectiveness of the system were unreliable due to the lack of user feedback and the absence of actual user interaction logs; therefore, a qualitative prototype-based evaluation was performed, which demonstrated that the system was capable of delivering personalised, diverse, and sentiment-consistent recommendations. Contrary to many existing hybrid recommender systems, the proposed framework is a combination of the theoretical and experimental aspects of the work, thereby leading to a substantial step towards the practical deployment of a viable model.