Although true understanding of the meaning of natural language by computers is still a long way off, a first estimation of the semantics of words can be derived by looking at co-occurring words. Two applications developed at the ErasmusMC make use of this notion to help biomedical researchers: Anni first relates words to more generic concepts, and summarizes frequently co-occurring concepts in a concept profile. Comparison of the concept profiles can be used to predict biological relations, and have been shown useful in the interpretation of high-throughput experiment results. Jane creates word profiles of documents related to authors and journals, allowing authors of new scientific papers to find suitable journals for publication, and allowing editors to find suitable reviewers.
Presentation at SSR