Abstract. Linked Data makes available a vast amount of data on the Semantic Web for agents, both human and software, to consume. Linked Data datasets are made available with different ontologies, even when their domains overlap. The interoperability problem that rises when one needs to consume and combine two or more of such datasets to develop a Linked Data application or mashup is still an important challenge. Ontology-matching techniques help overcome this problem. The process, however, often relies on knowledge engineers to carry out the tasks as they have expertise in ontologies and semantic technologies. It is reasonable to assume that knowledge engineers should require help from the domain experts, end users, etc. to contribute in the validation of the results and help distilling ontology mappings from these correspondences. However, the current design for the ontology-mapping tools does not take into consideration the different types of users expected to be involved in the creation of Linked Data applications or mashups. In this paper, we identify the different users and their roles in the mapping involved in the context of developing Linked Data mashups and propose a collaborative mapping method in which we prescribe where collaboration between the different stakeholders could, and should, take place. In addition, we propose a tool architecture based on bringing together an adaptive interface, mapping services, workflow services and agreement services that will ease the collaboration between the different stakeholders. This output will be used in an ongoing study to constructing a collaborative mapping platform.