System and Method for Creating and Searching Medical Ontologies

A method for creating and searching medical ontologies comprising the steps of:

• submitting a query to a plurality of ontologies, wherein each query comprises an operator expression, wherein each ontology returns one or more rooted directed acyclic graphs as a result of said query, wherein nodes of said graph represents concepts and edges of said graph represent links between concepts;
• calculating a similarity of each pair of graphs, and retaining those graphs whose similarity is greater than a predetermined threshold;
• for each pair of graphs g1 and g2 having a common subgraph, calculating tuple similarities for all pairs of tuples t1 and t2, where t1εg1 and t2εg2;
• for those tuples whose similarity is greater than a predetermined threshold, boosting a confidence of the tuples;
• and
• merging those graphs whose merger would yield a greatest confidence boost by adding edges not included in the common subgraph, to form g1∪g2.

A program storage device readable by a computer, tangibly embodying a program of instructions executable by the computer to perform the method steps for creating and searching medical ontologies, said method comprising the steps of:

• providing a semi-structured information source comprising a plurality of articles linked to each other, each article having one or more sections, wherein each article is associated with a concept;
• creating a directed unlabeled graph representative of said information source, wherein each graph node is associated with a concept and each graph edge is associated with a link between concepts;
• providing a plurality of labels, and labeling a subset of said edges of said graph;
• assigning each unlabeled edge an equal probability of being assigned one of said plurality of labels;
• updating, for each node, the probability of each outgoing edge by smoothing each said probability by an overall probability distribution of labels over all outgoing edges of each node;
• updating, for each node, the probability of each incoming edge by smoothing each said probability by an overall probability distribution of labels over all incoming edges of each node;
• and
• assigning a label with a maximum probability to an edge if said maximum probability is greater than a predetermined threshold to create a labeled graph, wherein said labeled graph represents a new ontology adapted to being searched.