Robotic Process Automation (RPA), the automation of office processes by software robots, promises companies improvements in efficiency and cost savings: While classical automation systems require profound changes to the used software products, the software robots mimic a human interaction with the graphical user interfaces of the applications. This allows to automatically control any functions of the software without the need to access the underlying implementations via separate programming interfaces (APIs).
However, existing RPA solutions have the problem that each process to be automated must first be processed manually by a domain expert and the necessary process knowledge must be populated into the system. This requires that the process has been neatly documented or at least is known by the expert. In practice, however, this is often not the case. The detailed recording and documentation of the existing processes as well as their translation into clear execution rules means that the configuration of a software robot today still requires a lot of manual effort.
With Desktop Activity Mining, the August-Wilhelm Scheer Institute is developing a method with which office processes can be automatically recorded and documented. This eliminates the previously necessary preparatory work for RPA and the full potential of process automation can be tapped.
HOW IT WORKS
The basic idea behind Desktop Activity Mining is to capture the screen actions of all involved employees, to consolidate and to merge them into a business process (Image 1). For this purpose, techniques from data and process mining are used. In detail, the relevant process actions of the employees, such as text input, mouse clicks or program calls, are identified and anonymized via a recording program running in the background. From these, process models are generated with process mining algorithms in order to obtain a comprehensive representation of the real work process.
Unlike traditional process mining approaches, where only transaction data from IT systems is used to describe a business process, Desktop Activity Mining relies on the direct desktop level. Therefore, even those actions are recorded that, despite their process relevance, cannot yet be documented by existing IT systems – for example, when writing an e-mail. Desktop Activity Mining is therefore the first approach that can holistically map processes on the actual work level of employees.
Image 1: Desktop Activity Mining captures human interactions with graphical user interfaces, consolidates data, and automatically creates detailed process documentation using process mining methods. This serves, for example, as the basis for the configuration of software robots.
The range of services offered by the AWS-Institute in the field of desktop activity mining covers all necessary steps for process entry and documentation via a staggered process: