Information extraction across multiple expertise-specific subject areas

A computer-implemented method for bridging terminology differences between at least two subject areas, comprising executing the following steps on a computer:

• obtaining a first corpus associated with a first subject area and a second corpus associated with a second subject area;
• for each obtained corpus, performing the steps of:
• computing a glossary;
• computing an affinity matrix between pairs of terms in the glossary and assigning scores according to a similarity measure;
• and
• computing a transitive closure of the affinity matrix and assigning a score for a pair of terms in the transitive closure of the affinity matrix using a composite path probability;
• computing a set of bridge terms by intersecting the respective glossaries of the first corpus and the second corpus;
• and
• computing a synonym dictionary as a set of triples S(f, t, w) where f is a term in the glossary of the first corpus, t is a term in the glossary of the second corpus, and there exists a term b in the set of bridge terms such that a term triple (f, b, t) is in a join of the transitive closure of the affinity matrix of the first corpus, the set of bridge terms, and the transitive closure of the affinity matrix of the second corpus, and where w is the composite path probability of (f, b, t),
• wherein the obtaining and computing steps are performed by a computer processor.