Integration of Explicit and Implicit User's Feedback with Contextual Information in Hybrid Recommender System

Recommender system helps users identify relevant content from the growing volume of information available on digital platforms. However, user preferences are dynamic and may not be adequately represented by a single source of feedback. This study proposes a hybrid movie recommendation system that integrates explicit feedback, implicit behavioural feedback, and temporal contextual information Using a convolutional neural network (CNN). Explicit feedback is represented by user’s movie ratings, while implicit feedbacks represented by interaction frequency. Contextual information is derived from users’ interaction timestamp to capturer changes in preference over time. The movie lens 1m is used as the main source dataset, from which 10,000 rating records were selected for the experiment. The selected records were pre-processed to construct the three information sources and used to train and evaluate the proposed model. Performance was accessed using precision, Recall, F1-score and mean squared error (MSE). The model achieved a precision of 86.25%. Recall of 98.76%, F1-score of 92.08%, and MSE of 0.0845. The results indicate that the Proposed architecture can integrate explicit feedback, implicit feedback and contextual information within a unified recommendation framework. The study demonstrates the feasibility of combining these information sources for personalized movie recommendation. The findings provide a basis for developing recommendation systems capable of delivering more relevant and personalized content in dynamic digital environments.