Abstract: A method, system, and computer readable code for improving user exploration (i.e. navigation and browsing) through hierarchies of information. Entity relevance is calculated for the entities in the hierarchies, and this information is used to guide the user in his exploration. Two types of entity relevance are used, structural relevance and state relevance. Structural relevance is preferably pre-computed, being static in nature. State relevance is dynamically computed, as the user selects nodes or categories during exploration. Each selection serves to further constrain the solution space. Entities are determined to be strongly relevant, weakly relevant, or irrelevant for the current solution space. Node feasibility is also dynamically calculated, and provides a forward-looking indication of which nodes can be selected while still remaining consistent with the current solution space.