Recent improvements of the BEL Information Extraction workFlow (BELIEF) for biomedical text mining and curation

Construction of structured knowledge requires technology that links text mining and curation to knowledge repository. We recently presented BEL Information Extraction workFlow (BELIEF) as a tool that facilitates the transformation of unstructured information described in the literature into structured knowledge networks. BELIEF automatically captures causal molecular relationships from scientific text and encodes them in BEL statements. BEL (Biological Expression Language) is a computable and human readable language for representing, integrating, storing, and exchanging biological knowledge in causal and non-causal triples. Recently, we have improved the curation process by extending the biomedical vocabulary and by making the curation dashboard more flexible. Moreover, BELIEF was enhanced with the integration of the OpenBEL API that allows direct linkage to the OpenBEL platform and enables upload of curated documents into the BEL knowledge base. These technological developments of BELIEF greatly improve the curation process and make the BEL knowledge more manageable. We continually use the BELIEF to develop an extensively annotated knowledge base of BEL triples that serve as building blocks for causal biological network models.