Multi-Language Descriptive Text And Speech Generation Of BPMN Graph

With the popularity of Internet,more and more companies and organizations use process models to organize and document processes.Process model has some unique features,such as complex semantics,diverse executive departments and model elements,thus participants have difficulty in reading the process model.The descriptive text of process model is more intuitive and understandable than formal process model.However,the existing methods generate descriptive text from process model instead of process graph.In addition,only single language descriptive text of the process model is generated.In general,participants have different cultural backgrounds,and mutiple language descriptive texts of the process model are necessary.Multi-language descriptive text and speech of BPMN graph are generated in this thesis.Main contributions are as follows:1.An automatic method is proposed to identify the BPMN model from BPMN graph.There are four steps:(1)BPMN graph elements are identified based on the template matching;(2)BPMN graph is decomposed based on node location and size,then the node text is identified by OCR technology;(3)The directed edges are identified based on the BPMN graph gray value;(4)The BPMN file is generated based on the identify results.2.Multi-language descriptive text is generated based on annotated process structure tree of BPMN process model,which includes three aspects:(1)The text information is parsed by semantic analysis and translation tools;(2)The process structure is divided into process fragments,and the process structure tree is constructed based on the process fragments.Then,the parsed text information is added to the process structure tree to generate the annotated process structure tree;(3)The multi-language descriptive text is generated based on the annotated process structure tree.3.A generation method of multi-language speech from BPMN process model is proposed.According to the structure complexity and the text information,annotated process structure tree is decomposed into several subtrees.The multi-language segmented texts are generated from the annotated process structure subtrees,and multi-language speech is generated by text to speech technology.4.The generation system of multi-language descriptive text and speech is designed and developed,then the proposed methods are verified in this system.