Generally robots need to use maps while navigating from one place to another. Metric (grid-based) and topological (node-based) paradigms are utilized to produce maps. In this work, topological map of a large-scale indoor environment is constructed from its metric map. To do this, we follow three stages: 1) Construction of metric map; 2) Determination of nodes; 3) Connecting the nodes to produce the topological map. Apart from the studies in literature, only the new cells in the metric map are considered to determine the nodes. As a result, the topological map grows in an online manner and computational cost of spectral clustering and extended Voronoi graph is reduced. The methods are tested in a simulated model of ESOGU Electrical Engineering Laboratory building in Gazebo simulation environment by using ROS.