GraphNetNLP-Enhanced Graph Recurrent Mapping Framework for Social Computing Recommendation

The growing complexity of service demands and user interactions in the Internet of Everything (IoE) necessitates intelligent systems that can adapt and comprehend human interactions. The interactions are interpreted through multiple modalities, including audio, text, and real-time observations, enabling personalized and accurate responses via Natural Language Processing (NLP) techniques. The existing approaches struggle to manage the uncertainty and computation delay while processing voice inputs. Therefore, this paper introduces Graph-Aided flow-based neural networks for NLP (GraphNetNLP) to effectively analyze and leverage user intent for intelligent social computing recommendations. The GraphNetNLP approach integrates the graph networks and natural language processing techniques that utilize voice and audio inputs to develop an interconnected cyclic graph that can be used to understand the demands and services provided to the user. In addition, recurrent graph mapping and training processes are utilized along with the certainty factors to analyze the disconnections. The certainty factors are selected based on the mapped and exhausted features of each iteration. The successful training process minimizes computational complexity by up to 8.86%, uncertainty by up to 6.87%, improves accuracy by up to 9.68%, and increases processing ratio by up to 11.77%. Thus, the introduced GraphNetNLP-based information processing is effectively utilized in text and other (IoE) applications.