Authors:Leonardo Guerreiro Azevedo1;Raphael de Almeida Rodrigues2 andKate Revoredo2
Affiliations:1Federal University of the State of Rio de Janeiro (UNIRIO) and IBM, Brazil;2Federal University of the State of Rio de Janeiro (UNIRIO), Brazil
Keyword(s):BPM, BPMN, Natural Language Generation, Natural Language Processing.
RelatedOntology Subjects/Areas/Topics:Artificial Intelligence and Decision Support Systems ;Enterprise Information Systems ;Industrial Applications of Artificial Intelligence ;Natural Language Interfaces to Intelligent Systems
Abstract:The proper representation of business processes is important for its execution and understanding. BPMN is the de facto standard notation for business process modeling. However domain specialists, which are experts in the business, do not have necessarily the modeling skills to easily read a BPMN model. Natural language is easier for them to read. Thus, both model and text are necessary artifacts for a broad communication. The manual edition of both artifacts may result in inconsistencies, due to unilateral modifications. This research proposes a framework for synchronizing BPMN model artifacts and its natural language text representation. It generates textual work instructions from the model, and it updates the original model if the textual instructions are edited. The framework was implemented using Java standard technology and evaluated through experiments. The first experiment concluded the textual work instructions can be considered equivalent to process models in terms of knowledge representation. The second experiment concluded the knowledge represented by the manually updated text can be considered equivalent to the automatically updated process model after the synchronization.(More)
The proper representation of business processes is important for its execution and understanding. BPMN is the de facto standard notation for business process modeling. However domain specialists, which are experts in the business, do not have necessarily the modeling skills to easily read a BPMN model. Natural language is easier for them to read. Thus, both model and text are necessary artifacts for a broad communication. The manual edition of both artifacts may result in inconsistencies, due to unilateral modifications. This research proposes a framework for synchronizing BPMN model artifacts and its natural language text representation. It generates textual work instructions from the model, and it updates the original model if the textual instructions are edited. The framework was implemented using Java standard technology and evaluated through experiments. The first experiment concluded the textual work instructions can be considered equivalent to process models in terms of knowledge representation. The second experiment concluded the knowledge represented by the manually updated text can be considered equivalent to the automatically updated process model after the synchronization.