Digital Business Processes

The Digital Business Processes research group has many years of experience with digital infrastructure data and creating added value by data processing. The group’s expertise is based on Big Data and includes various approaches to data mining and machine learning, which is one of the main aspects of Artificial Intelligence (AI). Thus, the group covers the entire value chain, starting with the processing of raw data and ending with the development of monetization strategies for digital services. Digital infrastructures and Industry 4.0 are two of the domain-specific fields in which the research group pools their scientific expertise and practical experience. The institute’s own Big Data Cluster, which makes intelligent use of both software and hardware components, provides the technological foundation for the fast processing of data as required by industry and science.  

Artificial Intelligence has left the research labs, and with breathtaking speed, it is starting to permeate our daily lives in the shape of digital assistants, cooperative robots and autonomous vehicles. The following publication offers an interesting overview of the current state in the transformation process towards digital economy:


Start of project in February, 2019

The project AMCOCS (Additive Manufactured Component Certification Services) includes the development of a self-learning platform to accelerate testing and certification procedures in additive manufacturing and the transfer of these findings into digital business models.

The use of additive manufacturing requires the provision of durability approval. Currently there are only very time- and cost-intensive quality assurance processes for additive manufacturing, which represent a high hurdle for the broad application of this method. The research project created test platform will collect all data from additive manufacturing, post-processing and material parameter determination. As part of this project, AMCOCS will be piloted with use cases from the aviation sector. These investigations, the associated test and verification procedures as well as historical data from relevant tests comprise the data basis. Additional data is collected for each further test, which can be used to permanently refine the procedures. Complex algorithms can be used to predict the durability of the objects to be printed. In future, it should be possible to make reliable statements about the durability of the objects to be printed in advance. Based on the determined material properties, users will be able to make a professional interpretation. This significantly shortens manual testing and certification procedures.


Start of project in August, 2018

The DatenTanken (DataRecharging) project develops and defines digital business processes concerning the economically viable operation of mobility and energy services and transfers the results to digital business models. This way, the grid-compatible construction of both the public charging infrastructure for electric vehicles in Dresden as well as the partly public infrastructure for logistics providers and residential building companies is promoted.

It is planned to construct a total of 33 fast-charging stations and 237 regular charging stations in Dresden, Saxony’s capital city. At full capacity, a charging infrastructure such as this is able to reduce NOx emissions by 17.9 tons per year. In order to allow an economical operation of the infrastructure in the long term, the Fraunhofer IVI additionally designs innovative, databased business models within DatenTanken: During the recharging process, car owners have the option to grant access to their vehicles’ data in exchange for a reduced electricity price. This data is then either used for digital mobility services and/or sold to third parties, thus refinancing not only the charging infrastructure but also the reduced electricity price.


Automatic Recognition of Invoicing

Dynamic, self-learning machine learning tool for accounting in tourism

The aim of AuReBu is to detect invoices through methods of machine leaning and to integrate the results into a value-driven workflow. The steps are divided into

  1. Detection of business transactions
  2. Account allocation and assignment.

A data set consisting of 300,000 documents was used in the training phase of the AI components. In the process, rule sets, such as accounting areas, were defined by the development partner - the world's leading tour operator company. These rule sets were described so that it was possible to establish dynamic rule development within a self-learning system. It is important to mention that at a certain point the AI system entered the self-learning phase so it was able to develop a self-adaptive rule system. At that moment, pre-defined rules were a thing of the past.

Within the prototype, the following components were developed, which in turn consist of over 40 sub-components: 

  • Document analysis 
  • Machine learning/AI 
  • Integration of domain knowledge 
  • Optimization of the workflow

At the end of the training phase, mass comparisons were conducted between the accounting records made by the Aurebu AI and those made manually by humans. On average, the Aurebu AI achieved a matching rate of >85 %. The tests have also shown that the AI system can speed up the processing of accounting by a factor of 10. The only limitation is the use of scalable hardware components. Additionally the use of training data in the form of current invoices should automatically increase the matching rate further.


Cartox² develops and defines digital business processes concerning basic services of autonomous driving, and converts its findings into digital business models. Their services are directed at enterprises and public authorities that are going to provide services for autonomous driving in the future (automobile manufacturers and component suppliers, telecommunications companies, as well as digital mobility service providers). The basis for these processes is formed by digital map material and additional information for automated driving. Crucially interesting information for the target group comprises information on car-2-car connectivity, grid coverage for car-2-infrastructure communication, as well as an analysis of the data routing to and between the edge clouds, the access points and the cloud environment. Based on this, Cartox² offers analysis results for low-latency data processing in real-time, such as risk assessments for traffic participants. In the process, data from various sources, such as the broadband atlas or 3D city models, and a large number of proprietary, hybrid communication and localization data are integrated. These are compiled on a Big Data platform, enhanced and transmitted to three initial Cartox² services. This way, detailed data acquisition can be avoided and the digital services can be obtained "as-a-service".


Intelligent Traveller Early Situation Awareness

Within the iTESA project, an automatic, real-time alert system for travel risks is created with the help of machine learning methods. This system alerts travelers in real-time, thus making it possible to change arrangements even immediately prior to departure. The primary trigger for iTESA’s acquisition of the corresponding information is social media. Public sources then serve to validate the information. The resulting data stream is analyzed in terms of travel risks (riots, natural disasters, epidemics) with the help of semantic dictionaries and statistical methods. The basis for this is the expertise of Europe’s  leading travel data platform and the historical crisis management log of world's leading tour operator company. iTESA enables travel agents to offer a valuable additional service that is of especially high interest to enterprises, as they carry the legal duty to care for the welfare of their employees. Thanks to their interoperability and generic approach, the techniques and methods developed within iTESA can also be transferred to other sectors, such as logistics.

Project Website


Internet Social Analyser in Natural Disasters

Within the scope of iSAND, an automatic information system for real-time risks in disaster situations was devloped. The results were implemented in a digital workflow management system with the aim of developing measures for the adaptation to disaster situations and at increasing the adaptability of interactive social systems. Its main focus is the processing of information shared in social networks and the laying of foundations for steering large groups of people towards orderly and sensible help measures. To achieve this, the data in critical areas is processed with the help of machine learning methods using data from social media. The results are transmitted and made available as prototypes in a live system.