In this thesis, we investigated the compression of directed graphs, which are used to model different types of networks such as social networks, biological networks, etc. As graphs grow in size, problems arise in their storage and processing. We implemented two methods that compress directed graphs using symmetries, which were calculated using graph automorphism. We then combined them with lossless compression methods and compared them with existing directed graph compression methods. The compression quality was evaluated by the number of bits per edge. We evaluated the performance of approaches on a dataset of graphs from various domains. The methods using symmetries were not more successful than the existing ones, but we proposed many possible improvements.
|
