Enhancing Declarative Temporal Model Mining in Relational Databases: A Preliminary Study

Appleby, S; Bergami, G; Morgan, G

doi:10.1145/3589462.3589491

Enhancing Declarative Temporal Model Mining in Relational Databases: A Preliminary Study

Lookup NU author(s): Sam Appleby, Dr Giacomo Bergami ORCiD, Professor Graham Morgan ORCiD

Downloads

Published version [.pdf]

Licence

This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0).

Abstract

Propositionalisation tampers the running time of state-of-the-art algorithms in declarative temporal model mining, as they exhaustively generate the clauses instantiated with the results of frequent itemset mining algorithms. Existing algorithms also exploit non-indexed data representations, thus negatively affecting the overall running time. This paper proposes a novel temporal model mining algorithm, Bolt, twinning confidence and support metrics as heuristics for candidate pruning with data structures enabling fast temporal data scanning. Bolt outperforms both state-of-the-art and renditions of existing mining algorithms using KnoBAB as a library