Security monitoring is one subdomain of cybersecurity which aims to guarantee the safety of systems, continuously monitoring unusual events. The development of Internet Of Things leads to huge amounts of information, being heterogeneous and requiring to be efficiently managed. Cloud Computing provides software and hardware resources for large scale data management. However, performances for sequences of on-line queries on long term historical data may be not compatible with the emergency security monitoring. This work aims to address this problem by proposing a semantic caching framework and its application to acceleration hardware with FPGA for fast-and accurate-enough logs processing for various data stores and execution engines.