Hybrid AI Pipelines
- Presentation
- Artificial Intelligence (AI) in Technical Communication
- 12. November
- 11:15 AM (CET) - 12:00 PM (CET)
- C5.2/3
-
George Bina
- Syncro Soft / Oxygen XML Editor
Contents
The recent evolution of AI enables many interesting processing workflows that used to be considered very challenging, if not impossible. However, AI processing alone is not optimal for solving real world problems. A hybrid processing that mixes AI processing with traditional data processing provides the best approach to tackle various information processing tasks.
XProc is a W3C recommendation, an XML Pipeline language that can orchestrate various workflows just by connecting a set of processing steps, so it is ideal for traditional structured data processing. We experimented with adding an AI processing step that allows interaction with an AI engine, thus enabling hybrid processing that mixes AI and traditional information processing. We then applied this to some real world use cases proving that this innovative approach of mixing AI and traditional processing in such a simple yet powerful framework is very useful in solving real-world problems.
Discover hybrid AI processing, the use cases we experimented with, how easy it is to define your own workflows, and how you can benefit from mixing AI and traditional processing to find solutions to difficult problems.
Takeaways
Hybrid AI processing allows to decompose a complex problem in multiple simpler tasks and these tasks can be orchestrated easily with XProc to split and combine various data flows, and to mix traditional and AI processing steps.
Prior knowledge
Awareness of structured content, XML in particular, will be useful, but it is not a hard requirement.