A virtual assistant for supporting knowledge workers in the execution of routine work

In this project, a workflow management system was combined with a chatterbot interface and a recommender system in an attempt to create a virtual assistant that can assist knowledge workers during the execution of a business process. A recommender system was built using different components such as neural networks for predicting workload and case difficulty, an external text analytics system for performing sentiment analysis, as well as a natural language interpreter used for processing inputs from the user in natural language through the chat interface. The developed artifact was tested with two groups of test subjects in order to determine if the assistant would increase the performance of users when executing a business process using a virtual assistant, versus users not using it. Although there was not a clear indicator of a performance difference for the two groups, insights regarding the expectations of test users for a virtual assistant were recorded, and the evaluation of an updated assistant including these additional features is left as a topic for further research.