Social Media Link Prediction For Friend Recommendation Using Graph Mining
Keywords:Graph mining, Exploratory data analysis, Random Forest, NetworkX, precisiom@topk, Kartz Centrality, HITS Algorithm.
Social media plays a vital role in today's interconnected world, and establishes a nexus in various ways. It breaks down geographical barriers and enables individuals to interact with others regardless of their location. Moreover, it is an essential tool for businesses and professionals. It facilitates networking, brand promotion, and customer engagement. It also serves as a platform for job seekers and recruiters to connect. During crises such as natural disasters or emergencies, social media can be a lifeline for affected communities.
In recommendation systems, link prediction helps improve user experience by suggesting connections or content that users are likely to find relevant. This is widely used in platforms like Netflix, Amazon, and social media to suggest friends, products, or content. It is vital for understanding and analysing the dynamics of social networks and also identify potential connections, friendships, or collaborations that are likely to form in the future.Graph mining for link prediction is valuable in applications where understanding the potential relationships between entities is crucial.
In this research there is an overview of Graphs, node, vertex, edge, path, Data format & Limitations. Exploratory data analysis is used to help identify obvious errors, as well as for better understanding of patterns within the data, detect outliers or anomalous events and find relations among the variables. In addition, to get optimal recommendations Feature engineering is used on Graphs with similarity measures including Jaccard & Cosine Similarities, PageRank, Shortest Path, Connected components, Adar index, Kartz Centrality, HITS Score, Weight features. Finally missing links are predicted by training a model.