LTS: Learning to Search

LTS (Learning to Search) is an implementation of an algorithm described in "LTS: Discriminative Subgraph Mining by Learning from Search History" in Data Engineering (ICDE), IEEE 27th International Conference, pages 207-218, 2011. The purpose of LTS is to find discriminative subgraphs, which are smaller graphs that are embedded in larger graphs that all share a certain trait. A discriminative subgraph can help to characterize a complex graph and can be used to classify new graphs with unknown traits. LTS is an improvement on other subgraph mining algorithms because it uses empirical data from the search history of a first pass to help weed out unpromising search directions. This allows a second pass to spend more time investigating promising areas of search to generate the most discriminative subgraphs more quickly than before.