Extracting existed knowledge output metadata belonging to institute members from other systems is an important way for building institutional repositories. This article analyses the methods of metadata sharing of some common knowledge resource management systems, provides three metadata acquisition strategies. Attempts to solve the problem of author disambiguation combining with technologies of semantic web. Finally, the system collects metadata form WOS for testing. The result shows the automatic metadata acquisition method based on the researcher ontology can do the greatest possible to collect metadata form other systems, and which is a good solution for author disambiguation.