Enhanced Community Identification Based on Spider Ant Colony Relative Feature Mining Algorithm Using Dense Net Convolutional Neural Network

Community detection is essential for identifying related groups in online social networks. Internet Communication Technology (ICT) makes e-learning and technology necessary in education. In addition, ICT plays an important role, from providing distance education and identifying interest groups to providing services to social interaction groups. Also, social realities, relative adversaries, groups to be served, etc., may be determined based on vested interests. Therefore, it is necessary to enable active researchers in social networks and identify communities to discover more, mainly behaviour-based groups. However, not any comment can be copied in terms of social associations. It is demonstrated that most machine learning algorithms using absolute mean ratios can derive interest scores based on social group organizations. However, these can reduce accuracy due to random correlations and lead to poor analysis. To solve this problem, we can implement deep learning-based social recognition and improve recognition accuracy. An improved community identification based on Spider Ant Colony Relative Feature Mining Algorithm (SACR-FMA) using Dense Net Convolution Neural Network (DNCNN) can be proposed. The online Community dataset is collected based on learner search features. The Learners Impact Behavioural Rate (LIBR) is initially estimated to find the learner’s interest. Then Edge Graph Semantic Subset Group Mining (EGSSGM) is used to marginalize the feature rate. Depending on the learner community rate, the Spider Ant Colony Relative Feature Mining Algorithm (SACR-FMA). The selected features are trained into Dense Net Layers (DNL) to create relative subgroup identity mining. Then the features are introduced into an optimistic Dense Net Convolution Neural Network (DNCNN) to identify the community interest class by category. The proposed method performs very well in detecting related subgroups in social networks compared to other systems. It provides accuracy, higher precision, recall and F-measure with less time complexity.