Empirical research into the novice-expert differences in information requirement analysis has recognized that the differences in knowledge and in modeling behaviors are the causes of differences in quality of requirement specifications. However, there is no cognitive process model available for explaining the interactions among the three factors: knowledge, modeling behaviors, and the quality of requirement specifications. On the basis of structure-mapping model of analogy, this article proposes a cognitive process model that views information requirement analysis as a process of conceptual mapping from the base structures (i.e., the knowledge structures of requirement analysis techniques) to the target structures (i.e., the knowledge structures of users’ problem statements). Due to the differences in knowledge, novice and expert information analysts use different types of cognitive processes, relation mapping by experts versus object-attribute mapping by novices, to model information requirements. The different cognitive processes lead to different modeling behaviors, and in turn the different modeling behaviors finally result in different qualities of requirement specifications. On the basis of the cognitive process model, two ways to improve the performance of novice information analysts are suggested: encouraging novice information analysts to think in terms of relations rather than objectattributes and providing domain-specific requirement analysis techniques that are similar to the problem domains in both relations and object-attributes.