Building ontology is one of key issues for knowledge sharing which is important technology to realize largescale knowledge-base systems. In this paper, we propose a computational framework for constructive ontologies called aspect. We first formalize aspect in a logical framework and then define it as a programming language. Aspect is representation of conceptualization which consists of a vocabulary and a theory, and ontologies are built as composition of aspects. Two types of composition of aspects are provided, one is combination of aspects which is just union of aspects, and the other is category of aspects which links different but domainsharing aspects. Using them, we can represent not only relations among different aspects but also a set of aspects which either of them can be used if needed. We show a logical formalization of aspect by using modal logic. Category aspect is modeled using modal operator O. We also formalize characteristics for aspects like compatible and rigid which can be used as criteria for ontology. We also provide ASPECTO1, a programming language for aspect by extending Ontolingua. We then show translation of messages as a way of interpreting multiple aspects. A translation agent can translate a message with some aspect to one with another aspect by analyzing dependency of aspects. Mediation and translation of messages are important to build agents easily and naturally because less knowledge on other agents is requested for each agent.