It has been widely argued across e-learning research literature that the most innovative approaches to e-learning are those which allow learners to receive customized training material according to their needs and their own time and place. In this paper we propose an integrated approach which combines these constraints in generating a time-adjusted knowledge volume on-the-fly. Recent progress in semantic web could facilitate a real-time approach to learning whereby contextual instruction is delivered at a time where the learner is ready to receive it and within a preset duration which satisfies the learner's schedule. To achieve these objectives, we propose a rule-based representation to model a domain context. The related domain knowledge is encapsulated within standard learning objects. A progressive inference engine maps selected ontological entities into timed learning objects which temporal attributes are deemed feasible with respect to the learner's schedule.