Seq2Pat: Sequence‐to‐pattern generation to bridge pattern mining with machine learning

• Published in: The AI magazine Vol. 44; no. 1; pp. 54 - 66
• Main Authors: Kadıoğlu, Serdar, Wang, Xin, Hosseininasab, Amin, van Hoeve, Willem‐Jan
• Format: Journal Article
• Language: English
• Published: La Canada John Wiley & Sons, Inc 01.03.2023
• Subjects: Algorithms; Artificial intelligence; Case studies; Constraints; Data mining; Knowledge discovery; Libraries; Machine learning; Pattern analysis; Pattern generation; Supervised learning
• ISSN: 0738-4602; 2371-9621
• DOI: 10.1002/aaai.12081

Pattern mining is an essential part of knowledge discovery and data analytics. It is a powerful paradigm, especially when combined with constraint reasoning. In this overview, we showcase Seq2Pat, a constraint‐based sequential pattern mining (SPM) tool with a high‐level declarative user interface. The library finds frequent patterns in large sequence databases subject to constraints. We highlight key benefits especially desirable in industrial settings where scalability, explainability, rapid experimentation, reusability, and reproducibility are of great practical interest. We then bridge SPM with supervised machine learning via dichotomic pattern mining (DPM). DPM exploits the dichotomy between outcomes correlated with patterns that uniquely distinguish them. Last, we present an automated feature extraction powered by Seq2Pat and DPM to discover high‐level insights and boost downstream machine learning models for intent prediction in digital behavior analysis.