AI Innovation Challenge: Three Projects at the University of Mannheim Are Funded

As part of the “KI-Innovationswettbewerb” (AI innovation challenge), the Land of Baden-Württemberg funds 44 projects of middle-sized companies with 11.4 million euros in total. The funding aims at overcoming technological hurdles when commercializing artificial intelligence (AI) and at helping small and medium-sized companies to benefit from the great potential AI holds. The University of Mannheim is the research partner of three of these projects.

Virtual construction planning
The goal of the project “KI4VPB – Virtuelles Planungsbüro zur KI-gestützten Erfassung und Plausibilitätsprüfung von Bauplänen“ (KI4VPB – virtual planning office for AI-based registering and plausibility checking of construction plans) is to create a platform which digitizes the essential processes of construction planning and simplifies them by using AI. This way, the coordination efforts for projects with a large volume could be reduced which would shorten the processing time. This would also allow for an earlier detection of planning mistakes. The process is currently developed at the chair of the computer scientist Prof. Dr. Heiner Stuckenschmidt. A prototype will probably be available by September 2021.

Project partners: Institute for Enterprise Systems (InES) of the University of Mannheim, bwcon research gGmbH (Stuttgart), d-serv GmbH (Tübingen), Loidl GmbH (Göppingen) and The Chainless GmbH (Freiburg)

Multilingual language bots and language assistants across domains
“Conversational Artificial Intelligence (Conversational AI)” systems are automated dialogs via language bots and language assistants. These chat bots and language assistants are trained by using machine learning methods with texts and dialogs from the corresponding domains and languages. For example, they learn from real dialogs how to order food in Italian. However, the chat bot which has been trained for one domain cannot simply be transferred to another domain or work in other languages, since this would require training with dialogs in other domains and languages.

The project is headed by junior professor Dr. Goran Glavaš and analyzes the methods which can be used for transferring conversational AI models to other domains and languages, even if only a limited number or dialogs, or in the extreme case, no dialogs are available for the target domain or target language. Within the course of the project, the researchers will create AI agents for different domains and languages and provide them as “plug in” modules to researchers and companies.

Project partners: Institute of Computer Science and Business Informatics of the University of Mannheim, inovex GmbH (Pforzheim) and Neohelden GmbH (Karlsruhe)

Sally 2.0 – a social and intelligent chat bot
Sally is a chat bot which can answer questions on a product and present new products. Many companies use Sally as sales support. The goal of the project at the chair of Prof. Dr. Heiner Stuckenschmidt is to develop an optimized version of the chat bot which would allow Sally 2.0 to adjust to potential users. More precisely, the chat bot will be trained to create a user profile of the customer in real-time. Another goal is to optimize how the bot addresses users. The text analysis technology of the company 100W is the basis for further developing the bot.
Project partners: Institute for Enterprise Systems (InES) of the University of Mannheim, Aliru UG (Mannheim) and 100 Worte Sprachanalyse GmbH (Heilbronn)

Further information
All projects are sponsored as part of the AI innovation challenge by the ministry of economic affairs, labor and housing construction (Ministerium für Wirtschaft, Arbeit und Wohnungsbau) in Baden-Württemberg. The projects have started on 1 January 2021 and will run for 12 months.

For more information, please go to:
http://www.baden-wuerttemberg.de/de/service/presse/pressemitteilung/pid/land-foerdert-projekte-zur-kuenstlichen-intelligenz/

 

Back