For the interpretation of molecular patient data, it is becomes increasingly important to compare the personalized data directly with the current knowledge. This 'a priori' knowledge is usually only present in the current medical literature and is difficult to sift and extract without automatic support. The use of text mining methods helps to quickly get an overview of existing knowledge and, if necessary, extract relevant data and connections. This automatic support reduces the effort to create and maintain highly structured knowledge resources for the interpretation of patient data.
The development of methods and applications in biomedical text mining is one of Fraunhofer SCAI's research areas. An example is the semantic search engine SCAIView, which provides a quick overview of the Medline publications.
Another example is the information extraction workflow BELIEF, which supports the extraction of causal networks and additionally provides a curative interface for the evaluation and correction of the data.