Exploring the scope of explainable artificial intelligence in link prediction problem-an experimental study

The realm of SN has witnessed remarkable developments, capturing the attention of researchers who seek to process and analyze user data in order to extract meaningful insights for future predictions and recommendations. Among the challenging problems in SN analysis is LP, which leverages available data and network knowledge, including node characteristics and connecting edges, to forecast potential associations in the near future. LP is used in data mining, commercial and e-commerce recommendation systems, and expert systems. This research presents a thorough LP taxonomy, including Similarity Metrics and Learning-based approaches, and their recent expansion in numerous network environments. This article also discusses XAI, a method that helps people understand and trust ML systems. LP taxonomy based on XAI is also proposed. The research also examines LIME, a popular XAI approach that illuminates ML and DL models. LIME provides model-independent local explanations for regression and classification tasks on structured and unstructured data. The study includes an extensive experimental evaluation of incorporating XAI with LP, which shows the XAI approach's ability to solve LP problems and interpret predictions. This research uses XAI to give users practical insights and a better knowledge of the LP problem.