A Novel Negative Link Prediction Algorithm for Social Networks

In recent years, the use of social media networks has increased. With the rise in social network usage, social media businesses have begun to place a greater emphasis on large-scale social network analysis. As a result, several components of the signed network have been studied and analysed. Link prediction is a crucial aspect of this signed network analysis. The majority of previous study focused on forecasting the network's good aspects, such as favourable user connections, but the negative component(i.e., Negative Link) is just as significant as the positive (i.e., positive link). It enhances the performance of several current positive link analysis programmes. As a result, we concentrate on predicting negative ties from a social network that contains both positive and negative interactions in this research. To improve prediction precision, we use social theories such as balancing and status theory. We utilise the XGBoost classifier to determine if a particular missing link is positive or negative. Finally, we use accuracy and F1 measure measures to assess our proposed model.