The document proposes an ensemble technique that combines different unimodal recommender systems to generate recommendations based on multimodal user interactions. It uses matrix factorization and Bayesian personalized ranking as unimodal recommenders. An algorithm is presented that averages the scores from each unimodal recommender to produce a final recommendation list. The experiments show the proposed ensemble approach achieves better results than the baseline unimodal recommenders in terms of MAP and Precision metrics.