A New Approach to Minimize Memory Requirements of Frequent Subgraph Mining Algorithms

Frequent subgraph mining (FSM) is a subsection of graph domain which extensively used for classification and clustering purposes. Over the past decade, many efficient FSM algorithms have been developed. The improvements generally focus on reducing time complexity by changing algorithm structure or using parallel programming techniques. another problem to solve, high memory consumption. In this study, new approach called Predictive Dynamic Sized Structure Packing (PDSSP) proposed minimize requirement algorithms. Proposed redesigns internal data structures without any algorithmic modifications. PDSSP has two contributions. first one Integer Type (ds_Int) newly designed unsigned integer type. second contribution “Data packaging” component that uses packing technique changes behaviour compiler. A number experiments conducted examine effectiveness efficiency embedding it into state-of-art gSpan Gaston. implementation compared official one. Almost all results show consumes less each support level. As result, extensions can save peak usage may decrease up 38% depending dataset.