Due to the increasing amount of information in web-based environment, analysts nowadays need information extracted from different sources. Extracting this information to guide decision making in a national security perspective remains a challenging task. The major issue arises due to a large amount of irrelevant information or complexity of unstructured data which makes information extraction and classification a very tedious task while analyzing the textual content for finding various aspects of entities or groups. In this paper, we describe our efforts towards introducing a context-based information extraction using National Security Information Sources (NSIS) which employs different types of knowledge inspired by natural activities of living things. In order to improve classification performance by utilizing relevant information from the sources related to national security and to make better decision, we assume that entities and relationships from these sources can be used to contextualize information from the records. This paper presents a new method for nature-inspired classification after extracting various features from dataset created using social posts and records. Then, analysis of classes of information to concepts available on knowledge sources is carried out to ensure quality of information based on the user needs. The simulation results demonstrated the ability of the proposed method inspired by cultural algorithm to extract group-centric information and improve classification performance using proposed Social Info Finder (SIF). This paper highlights the challenges and application of proposed SIF method to improve information extraction and performance of text classification focusing on national security and related threats.