Content-integration systems generally suffer performance bottlenecks due to network overhead. To address this problem, the authors developed the Data Integration Cache Engine (DICE), which uses summarization techniques (subqueries and Bloom filters) and semantic metadata to achieve semantic active caching in the context of Internet-based data integration applications. The system uses algorithms and specialized data structures to generate exact remainder queries to locate content that is missing from cache in case of partial hits. The authors' performance results indicate that DICE outperforms existing options in terms of response time, network overhead, and server load.