Natural Language Analytics
Text is still one of the largest data types in almost any domain. Images and videos are becoming more and more popular, but the importance of text, both as means of communication as well as type of data companies work with, has not decreased. On the contrary, there are exciting trends happening in this field. One such trend are the so-called transformer models in natural language processing (NLP). A transformer in this context is a neural network that understands the context in text data and its meaning by tracking relationships in sequential data (e.g., words in a sentence). This means that we are steadily moving forward towards a state where machines will be able to not only understand text and speech the way humans do but will also be able to respond in a similar way. These language models are trained on very large datasets consisting of billions of parameters, and while training can take very long, it is possible to use pre-trained models for different tasks. This is part of another trend called transfer learning which allows businesses to complete various NLP tasks faster than before.
We use the latest developments in NLP from computer science and apply them in information systems research to solve real-world problems organizations face nowadays. An important part of our research is based on user-generated data in social media channels. We have been working with companies like Adidas, Kulmbacher, Datev on topics involving named entity recognition, sentiment analysis, and text summarization. We are also using social media data for understanding health misinformation spread, especially in German-language sources.
Researcher
Projects
- Anomaly correction driven by extracted expert knowledge and learning systems (ADELeS) (with Universität Augsburg, REHAU Industries SE & Co. KG, XITASO GmbH)
- Data-driven talent identification in sports marketing (with adidas AG)
- Social media analytics for customer insights (with Kulmbacher Brewery)