Erfolgreich eingeworben

BAICert – Bavarian Artificial Intelligence in Certification Processes: Reducing red tape for business, regulatory enforcement and certification through AI

How can artificial intelligence simplify document-intensive testing and certification processes? The ‘BAICert’ project is developing AI-supported methods to automatically analyse large volumes of different documents, process information and identify anomalies. The aim is to reduce bureaucratic burdens, ease the workload on specialist staff and make the solutions developed available for use by both the private sector and public administration.

What is the title of your research project?

BAICert – Bavarian Artificial Intelligence in Certification Processes: Reducing bureaucracy for business, public administration and certification through AI

Who is involved in the project?

Prof. Dr Jürgen Frikel (project lead); Prof. Dr Filippo Riccio; Catyana Heyne, M.Sc.

What is your funded project?

In the BAICert research project, we are developing AI methods, models and processing pipelines that can be used to automatically analyse and process large volumes of heterogeneous documents. These include, for example, methods for document classification, information extraction and structuring, as well as for the detection of anomalies and irregularities. Certification and auditing processes, in collaboration with our industry partner OmniCert, provide a challenging real-world application and testing ground for these technologies. The aim is to develop modular and transferable AI tools that can also be used in other document-intensive sectors of business and public administration.

What excites you about the project?

What is particularly exciting about BAICert is the close link between methodological AI research and a challenging real-world application scenario. From a scientific perspective, the processing of multimodal information is particularly appealing: relevant content is found in texts, images, tables or the visual structure of documents and must be captured collectively by AI models. Through our collaboration with OmniCert, such methods can be developed using real document collections and tested immediately under real-world conditions. A particular added value lies in automating repetitive manual tasks, such as the transfer and preparation of data, thereby reducing the workload on specialist staff. This leaves more time for the actual technical assessment, whilst document-intensive processes can be handled more efficiently at the same time.

How can insights from the project be applied in practice?

The AI assistance systems developed in the project can be deployed wherever large volumes of documents need to be captured, analysed and checked. They can automate repetitive manual tasks involved in the transfer and processing of information and, through anomaly detection methods, specifically highlight irregularities in documents and data. This reduces the workload on specialist staff, allowing them to focus more on the actual technical review and assessment. The methods and software modules developed are not limited to certification processes, but are intended to be transferable to numerous document-intensive areas of application in public administration, business and other sectors. Particular emphasis is placed on the development of open-source solutions that comply with data protection regulations; these are modular in design, can be flexibly adapted to different requirements and integrated into existing processes.

On our website, you will find an overview of OTH Regensburg’s current funded projects.