Abstract
We present a new approach to software engineering which reduces the knowledge gap between user and development methodology by explicitly supporting concepts expressed in natural language. The tool uses a natural language description of a business process as input and transforms it into a process model. The system recognizes actors, objects, locations, relationships etc. referred to in the description and distinguishes different types of actions and conditions. The system uses multi-pass parsing and disambiguation NLP techniques and relies upon a custom-built dictionary of 23.000 English root words. The dictionary includes information about syntactic (e.g. noun, verb...) and semantic categories as well as word frequency. Currently 15 different semantic categories such as 'tangible object', 'person', 'event', etc. are distinguished. The ACAPULCO prototype, which runs on a standard PC under Windows 3.1 with 16 Mbytes of RAM, demonstrates a) that natural language processing for software engineeringis feasible, b) that this approach has potential of redefining the interaction and relationships between users, analysts and developers and c) that this approach is a powerful extension to traditional methods because it uses explicit knowledge about real-world business concepts.
Recommended Citation
Hars, Alexander, "Advancing CASE Productivity by Using
Natural Language Processing and Computerized Ontologies:
The ACAPULCO system" (1996). AMCIS 1996 Proceedings. 56.
https://aisel.aisnet.org/amcis1996/56