Frank's Blog

The future of Modelling 

During the last year, we at inubit Labs investigated the possibilities to automatically create process models from natural language input, i.e. plain text. Just imagine the possibilities by combining this technology with voice recognition: While you interview a process participant (or even better: a c-level executive), a process model is automatically created during the interview. Combined with innovative process merge capabilities, you can even do workshops with many participants - and nobody has to do the drawing. Instead the systems shows you conflicts and missing information in real time. Another use case is the automatic investigation of existing texts, such as work instructions. Imagine a consultant walking into a new company and feeding the system with all the information available. He then gets a quick overview of duplicates, conflicts and missing information in a structured and formal way (e.g. process and other models). Even more, this information is stored in a "world model", that could be the foundation for a comparison with as-is processes captured via the methods discussed before.


You can figure out the actual state-of-the-art in a short video:

YouTube Link

Actually, it's quite a bit to go for the vision, but the core technology is yet available. What is still a manual task, is the post-processing of the created diagrams.

Further Reading

If you're interested in the topic, I recommend taking a look at our CAiSE'11 paper as well as the excellent Masters Thesis from Fabian Friedrich. And while we don't have a prototype online right now, our competition already has: Nevertheless, they require far more structured input than the approach we investigated, so that I doubt they will be able to handle the scenarios introduced above.


Add Comment

Comments are not available for this entry.