Recommender systems aim to assist web users to find only relevant information to their needs rather than an undifferentiated mass of information. Collaborative filtering (CF) techniques are probably the most popular and widely adopted techniques in recommender systems. Despite of their success in various applications, CF-based techniques still encounter two major limitations, namely sparsity and cold-start problems. More recently, semantic information of items has been successfully used in recommender systems to alleviate such problems. Moreover, the incorporation of multi-criteria ratings in recommender systems can help to produce more accurate recommendations. Thereby, in this paper, we propose a hybrid Multi-Criteria Semantic-enhanced CF (MC-SeCF) approach. The MC-SeCF approach integrates the enhanced MC item-based CF and the item-based semantic filtering approaches to alleviate current limitations of the item-based CF techniques. Experimental results demonstrate the effectiveness of the proposed MC-SeCF approach in terms of improving accuracy, as well as in dealing with very sparse data sets or cold-start items compared to benchmark item-based CF techniques.