Efficient Distance Pruning for Process Suffix Comparison in Prescriptive Process Monitoring

TL;DR

The paper addresses the heavy computational cost of suffix comparisons in prescriptive process monitoring, where outcomes depend on potential continuations. It introduces a pivot-based distance pruning framework that leverages the triangle inequality to bound $d(x,y)$ using distances to pivots $P=\{z_1,\dots,z_K\}$, enabling exact pruning. The approach uses a $k$-center objective for pivot selection and precomputes a distance matrix to accelerate queries, with batching to exploit parallelism. Empirical results demonstrate substantial speedups (e.g., from 89 hours to 2.5 hours) while preserving 100% accuracy relative to exhaustive search, supporting scalable prescriptive systems.

Abstract

Prescriptive process monitoring seeks to recommend actions that improve process outcomes by analyzing possible continuations of ongoing cases. A key obstacle is the heavy computational cost of large-scale suffix comparisons, which grows rapidly with log size. We propose an efficient retrieval method exploiting the triangle inequality: distances to a set of optimized pivots define bounds that prune redundant comparisons. This substantially reduces runtime and is fully parallelizable. Crucially, pruning is exact: the retrieved suffixes are identical to those from exhaustive comparison, thereby preserving accuracy. These results show that metric-based pruning can accelerate suffix comparison and support scalable prescriptive systems.