Constructing human brain-function association models from fMRI literature.

Toward the goal of understanding the human brain function, we have developed a web-based Human Brain Functional Mapping Knowledge Base (HBFMKB) system to mining human brain-function association model from vast Medline abstracts. Since nomenclature and relationships among cognitive functions have no consensus yet, we use rule-based natural language processing methods to extract behavioral task and cognitive function and do n-gram approximate concept mapping by the Unified Medical Language system (UMLS) knowledge source. The HBFMKB system has an automatic PubMed MEDLINE download and import system, name entity extraction system and interactive visualization system. In summary, the HBFMKB system helps scientists to get digest knowledge before design experiments and compare their results with current literature.