System and method for exploiting semantic annotations in executing keyword queries over a collection of text documents

A processor-implemented method of exploiting a plurality of semantic annotations in executing a keyword query over a collection of data, the method comprising:

• serializing the semantic annotations and storing them in an indexed storage device;
• generating strings by text analysis engines;
• generating a type system which models the semantic annotations, with the semantic annotations based on entity concepts and relationship concepts, wherein said entity concepts represent instances of data items that are mentioned at least one time in said collection of data and said relationship concepts represent associations between selected ones of said entity concepts and at least one other entity concept that represents another attribute of said selected entity concept;
• generating an interpretation index which indexes the semantic annotations against the type system, such that names of types, attribute names, and attribute values are indexed;
• expanding the keyword query into a plurality of intermediate structured queries, wherein one or more of keywords of the keyword query are matched by an interpreter with one or more entries in the interpretation index, utilizing matching techniques that retrieve approximate matches, and with the interpreter utilizing a shared ancestor rule which generates queries and discovers higher order relationships of keywords with matching types;
• wherein the intermediate queries include individual keywords and possible combinations of matches regarding each keyword, each of said intermediate queries representing separate queries having separate interpretations and said keyword query is represented by one of said plurality of intermediate queries, wherein the shared ancestor rule indicates that the keyword query contains keywords which have attributes that match each other with a specified type;
• modifying selected ones of the intermediate queries based on at least one relationship concept to generate additional intermediate queries, wherein keywords in said intermediate queries are replaced by corresponding relationship concepts;
• executing each of the plurality of intermediate queries to produce a list of data matching the keyword query of the executed intermediate query in terms of the type system;
• merging the list of data matching the intermediate queries to produce a single output list;
• and
• producing annotations regarding analyzed documents with a plurality of text analysis engines with off-line processing.