Learning object recommendation is an emerging technology providing learners with adaptive learning objects or personalized services to overcome the disorientation and overload problems due to massive information. This paper proposes a new ontology-based framework for semantic content recommendation towards learning process. Noting that learning is a continuous self-constructing process based on semantic activations of the learner's prior knowledge, the recommender includes three components: (i) ontology is used to represent the learning object content structure; (ii) semantic rules are developed to identify prerequisite concepts contributing to the understanding of the current learning object; (iii) concept lattices from the inferred concept sets are utilized to obtain the final recommendations. Incorporation of the contextual information as learning process and learning object content structure, this recommendation method is supposed to be effective, especially in the case of novices or learners with low-level prior knowledge, which is different with others.