OLAP Visual Analytics on Large Software Call Graphs with Hierarchical ChordMap

摘要

The performance of call graph analytics is critical to the development and management of large-scale software systems. Classical call graph visualizers either reuse a plain node-link graph metaphor that does not scale, or adopt the specific aggregation technique like the PivotGraph. In this paper, we first generalize the OLAP analysis framework from multidimensional data to multivariate graphs. Then we introduce Hierarchical ChordMap, a new visualization design that tightly couples with OLAP operations through context-preserving user interactions. Controlled user study demonstrates significant improvements of our design from the basic call graph visualizer. Implications are summarized on when and how the ChordMap design can outperform a stable PivotGraph tool in call graph analysis.