Dijkstra Algorithm Heuristic Approach for Large Graph

In this study, we developed an algorithm that improves a Dijkstra implementation in a large and dense graph based on hMetis partitioning. hMetis is a hyper graph partitioning algorithm that divides the massive graph into sub-graphs. The algorithm is called hDijkstra consists of pre-processing the flat graph by partitioning it onto a sub graph then building a two-level hierarchal graph. The shortest path algorithm computes on the toplevel graph. Our experimental results show the dominance of our algorithm over the traditional Dijkstra algorithm and other alternative solutions based on the time and the number of Input/Output (I/O) operations.