T-GOWler: Discovering Generalized Process Models Within Texts

Area of Science:

• Computer Science
• Information Systems

Background:

• Workflow management systems rely on explicit process models, which are difficult and time-consuming to create.
• Current workflow mining techniques face challenges with varying task abstraction levels in concrete workflow logs.
• Workflow logs contain ordered events that imply multilevel semantics and contexts, with potential for discovering generalized events.

Purpose of the Study:

• To propose an ontology-based workflow mining system for generating patterns from text-extracted event sequences.
• To address the challenges in creating explicit process models by automating pattern discovery.
• To enable the rediscovery of generalized events within complex processes.

Main Methods:

• Developed T-GOWler (Generalized Ontology-based WorkfLow minER within Texts), an ontology-based system.
• Utilized two core modules: a workflow extractor and a pattern miner.
• Employed two distinct ontologies: a domain ontology for text-based workflow extraction and a processual ontology for mining generalized patterns.

Main Results:

• Successfully demonstrated an ontology-based approach for extracting generalized workflow patterns from textual data.
• T-GOWler effectively mines patterns from sequences of events extracted from texts.
• The system leverages domain and processual ontologies to enhance workflow extraction and pattern mining.

Conclusions:

• Ontology-based workflow mining offers a viable solution for generating generalized patterns from text.
• T-GOWler provides an effective framework for discovering hidden semantics and contexts in workflow processes.
• This approach facilitates the creation of more abstract and generalized workflow models from unstructured textual data.