Organizations often have textual descriptions as a way to document their main processes. These descriptions are primarily used by the company's personnel to understand the processes, specially for those ones that cannot interpret formal descriptions like BPMN or Petri nets. In this paper we present a technique based on Natural Language Processing and a query language for tree-based patterns, that extracts annotations describing key process elements like actions, events, agents/patients, roles and control-flow relations. Annotated textual descriptions of processes are a good compromise between understandability (since at the end, it is just text), and behavior. Moreover, as it has been recently acknowledged, obtaining annotated textual descriptions of processes opens the door to unprecedented applications, like formal reasoning or simulation on the underlying described process. Applying our technique on several publicly available texts shows promising results in terms of precision and recall with respect to the state-of-the art approach for a similar task.

Textual descriptions of processes are ubiquous in organizations, so that documentation of the important processes can be accessible to anyone involved. Unfortunately, the value of this rich data source is hampered by the challenge of analyzing unstructured information. In this paper we propose a framework to overcome the current limitations on dealing with textual descriptions of processes. This framework considers extraction and analysis, and connects to process mining via simulation. The framework is grounded in the notion of annotated textual descriptions of processes, which represents a middle-ground between formalization and accessibility, and which accounts for different modeling styles, ranging from purely imperative to purely declarative. The contributions of this paper are implemented in several tools, and case studies are highlighted.

No abstract

The automatic extraction of formal process information from textual descriptions of processes is a challenging problem, but worth exploring, since it enables organizations to align complementary information that talks about processes. In this paper we continue our previous work on this area, based on defining hierarchical/tree patterns on the dependency trees that arise from the linguistic analysis. We now incorporate a new abstraction layer on these patterns, that consider relationships between nearby sentences. The aim of this extension is to capture inter-sentence relationships that typically arise in textual descriptions of processes. The experiments done on publicly available benchmarks corroborate this intuition, showing a significant rise in the ability to capture all the important control-flow relationships defined in the text.1 The reader can see a tutorial for annotating process modeling exercises in the ModelJudge platform at https://modeljudge.cs.upc.edu/modeljudge tutorial/.