Towards the Interactive Natural Language Explanation of Processes

Despite the efforts of the Process Mining community, the understanding of process mining results and reports by nontechnical users still remains an open challenge. For this reason, an increasing need for endowing, integrating and enriching process mining tools and pipelines with mechanisms able to convey the most salient aspects of a process in an understandable manner, arises. Within this context, natural modalities of communication, such as Natural Language, have proven as effective ways to convey information to users in an understandable manner in multiple domains. In this paper, we present a human-in-the loop approach for the interactive generation of Natural Language Explanations of Processes. The system allows non-technical users to interact, query and retrieve relevant and understandable natural language explanations of a process. Based in process mining algorithms and techniques, the proposed architecture employs fuzzy set theory for capturing the relevant semantics of a process within an ontology-driven system, capable of handling and reasoning with imprecise knowledge. We show the validity of our approach through the generation of a series of natural language explanations for a healthcare process, which have already been proven correct, relevant and understandable in the literature.

keywords: Process mining, Natural Language Generation, fuzzy quantification, Process Querying, ,