Image: AdobeStock
Research group We are developing advanced algorithms and models to find anomalies, patterns and similarities. This allows us to predict outcomes and facilitate decision-making even in very large and complex datasets.
The research group of Deep Data Mining was established to develop algorithms and implement prototype for multi-sources heterogeneous information federation and privacy preservation on multimodal data. Our main research interests include data federation and privacy protection by applying techniques for text mining, natural language processing, machine learning and semantic web.
Regarding datatypes to integrate, we consider data from structured (e.g, records in DB), semi-structured (e.g, XML, JSON) and unstructured sources (e.g, news, social media). In a broad view of the core techniques, our group applies technologies of database, data mining, natural language processing, machine learning, and ontology based semantic web technology.
As application-driven research, we aim to realize general data integration framework to adapt multiple applications (e.g, information retrieval, recommendation systems, online advertisements) and meanwhile acquire the unique characteristics of domain-data to boost the integration accuracy on specialized domains (e.g, social networks, demographic, review data).
Sonakshi Garg shows that privacy is not a barrier to progress.
"A remarkable evidence of our thriving research environment", says Frank Drewes, Head of the Department.
Christian Kammler studies how social simulations help decision-makers understand their impact on people.
