About BiographTA

Keywords: biomedical text mining, relation extraction, semantic processing, machine learning, domain adaptation


BiographTA is the part of the Biograph project carried out by the CLiPS research group.

Biograph is a project funded by the Bijzonder Onderzoeksfonds from the University of Antwerp (GOA BOF UA) that aims at putting forward a new methodology for text mining from heterogeneous information sources. The final goal of the project is to show new results in mining previously unknown relations between genes and diseases, and improved gene prioritisation discovering non-obvious disease causing genes. It is a multidisciplinary project within the University of Antwerp.

The main research goal within BiographTA is to develop supervised and semisupervised text mining techniques that allow to perform large scale biomedical relation extraction. The output of BiographTA will be integrated in the Biograph knowledge discovery server. This will allow to evaluate the newly developed techniques in real life applications.

The BiographTA tools extract from large corpora relations between biomedical entities weighted according to their reliability. Basic research has been performed to find new techniques that allow to calculate the reliability of relations based on processing discourse level linguistic information, such as negation, modality, and quantification. Exploring the efficiency and robustness of machine learning tools that perform deep text understanding is one of our main interests.

Additional research goals within BiographTA include exploring domain adaptation techniques in order to adapt existing text analysis tools to the biomedical language.

The BiographTA tools will be released for public use.

BiographTA participates in the biomina initiative, a new interdisciplinary research unit for biomedical informatics established by the University of Antwerp and the Antwerp University Hospital as part of the Clinical Research Center Antwerp (CRC Antwerp).