Ontology matching, aimed at finding semantically related entities from different ontologies, plays an important role in establishing interoperation among Semantic Web applications. Recently, many similarity measures have been proposed to explore the lexical, structural or semantic features of ontologies. However, a key problem is how to integrate various similarities automatically. In this paper, ... [Show full abstract] we define a novel metric termed a “differentor” to assess the probability that a similarity measure can find the one-to-one mappings between two ontologies at the entity level, and use it to integrate different similarity measures. The proposed approach can assign weights automatically to each pair of entities from different ontologies without any prior knowledge, and the aggregation task is accomplished based on these weights. The proposed approach has been tested on OAEI2010 benchmarks for evaluation. The experimental results show that the differentor can reflect the performance of individual similarity measures, and a differentor-based aggregation strategy outperforms other existing aggregation strategies.