The group researches how application-relevant data and information from heterogeneous sources (online and offline) can be systematically gathered, evaluated, processed and disseminated. The focus is on multimodal, semi- and unstructured data (e.g. text, audio and GIS data) from heterogeneous sources - i.e. resources that have not yet been systematically acquired and used. In order to utilize these, new concepts for filtering, contextualization, plausibility checking, validation and geolocalization are researched by combining and merging the multimodal data. The aim is to obtain processed, "analysis-ready" data and information from the heterogeneous raw data for subsequent analyzes and for decision support.
Basic and application-related methods and tools are developed to derive the properties of the resources (e.g. accessibility and quality) and the information they contain (e.g. correctness, content, spatial and temporal coverage). In addition, the systematic and robust derivation of relevant information is being researched. Central examples of basic tasks are the identification and the semantic, temporal or spatial aggregation of content-relevant information candidates. New concepts for the analysis, linking and fusion of data from heterogeneous sources form the basis for further plausibility checks, validation and geolocalization of these information candidates. A focus on specific sub-tasks (e.g. the binary classification of texts into “thematically relevant” or “irrelevant”) enables comprehensive empirical research of the methods, for example with regard to their transferability to other regions, languages or data, as well as their modular implementation.
The individual modules can be flexibly orchestrated into application-specific and comprehensible process chains, which makes it possible to quickly develop and provide initial solutions for new problems and applications.
Current application-oriented research topics: