Generative AI for the industry of tomorrow
FH JOANNEUM conducts intensive research in the field of generative AI; image © Adobe Stock
Project Manager RADIUS
Project Manager GenAI
Project Coordinator GenAI
With the GenAI project, a sub-project of RADIUS, FH JOANNEUM’s largest COIN project, FH JOANNEUM is researching the next generation of large language models. The aim is to make innovative approaches such as retrieval-augmented generation (RAG) and context-augmented language models usable in practice – for targeted knowledge queries, transparent context processing and personalised language models.
Large language models and generative AI
At the heart of GenAI are large language models (LLMs) that can understand, analyse and generate text. They form the basis for generative AI, which is capable of processing complex content in a context-sensitive manner and making it applicable for businesses and research.
Retrieval-augmented generation for more accurate answers
Retrieval-augmented generation (RAG) links the language model to additional data sources. As a result, the model not only provides generic answers, but can also retrieve and integrate context-specific information – a decisive advantage for industrial applications, e.g. in documentation or support systems, where targeted knowledge is crucial.
Context-augmented language models and NER
Another focus is on context-augmented language models, which interpret information not in isolation but in context. In combination with named entity recognition (NER), relevant entities – such as products, processes or actors – can be reliably identified and integrated into knowledge graphs.
Knowledge graphs as the basis for intelligent systems
The extracted information is used to create knowledge graphs that reveal complex relationships and can be integrated into real-time systems for information provision. GenAI thus creates the basis for smart applications that effectively provide targeted knowledge.
Personalised language models for practical use
The development of personalised language models tailored to specific requirements – for example, for specialist departments, companies or entire industries – is particularly exciting. This results in solutions that can be individually adapted and integrated into existing system landscapes.
With GenAI, FH JOANNEUM is demonstrating how generative AI can be designed in a practical manner and used in industrial contexts, making it a key project in this field.
