Summary: In a do-it-yourself software assemblage environment, it is important for the amateurs and technicians to find the right building blocks before assembling their own solutions. We have designed an ontology-based do-it-yourself architecture, which assists users to find suitable components and guide them to the assemblage. In particular, a tool called DIY-CDR (Do-It-Yourself Component Discover and Recommender) has been designed and implemented. The matching engine in DIY-CDR uses domain ontologies and annotation sets of the components and compares users’ requirements to the annotation sets. Since the components contain little metadata information and their descriptions are often free texts, how to automatically annotate these components becomes a problem. In this paper, we propose a solution called Onto-Ann, which is an automatic and semantically rich annotation tool. It uses combined technologies from natural language processing (NLP), social study and ontology.