Search query obfuscation via broadened subqueries and recombining

1. A method, comprising:

identifying, from a plurality of terms of a received query, a plurality of sensitive terms based on: (i) user input received with the query, (ii) an inverse document frequency (IDF) score of each term of the query, and (iii) a location of a respective node associated with each term of the query in an ontology;

referencing the ontology to identify a set of generalized terms corresponding to the plurality of sensitive terms of the query;

generating a plurality of subqueries, including a first subquery, a second subquery, and a third subquery based on the sensitive terms, a sensitivity score of each of the sensitive terms, and the set of generalized terms, wherein a first sensitive term of the plurality of sensitive terms is not one of the set of generalized terms and is excluded from each of the plurality of subqueries, wherein the third subquery is disassociated with the query, wherein the sensitivity scores are based on a predefined frequency threshold;

executing each of the plurality of subqueries to retrieve a result set for each respective subquery;

identifying, in a first result set corresponding to the first subquery, the first sensitive term;

determining that the first sensitive term is not present in a second result set corresponding to the second subquery;

applying a greater weight to the first result set corresponding to the first subquery relative to the second result set corresponding to the second subquery based on the first sensitive term being present in the first result set and not in the second result set; and

filtering the result sets using the plurality of terms of the query, including the plurality of sensitive terms, to produce a final result set responsive to the query.

2. The method of claim 1 , wherein the final result set includes at least one result from the first result set, wherein the greater weight applied to the first result set is further based on a publication source of at least one result in the first result set, wherein the publication source is not a search engine and corresponds to an entity publishing the at least one result, wherein each subquery includes at least one of: (i) at least one generalized term of the set of generalized terms, and (ii) a subset of terms of the at least one term of the query, wherein the plurality of subqueries are generated responsive to user input specifying to generate the subqueries, wherein the method further comprises:

merging the result sets after executing each of the plurality of subqueries.

3. The method of claim 2 , wherein at least two subqueries of the plurality of subqueries are executed on different search engines.

4. The method of claim 3 , wherein the set of generalized terms includes at least one broader term corresponding to the at least one term of the query.

5. The method of claim 4 , wherein generating the plurality of subqueries further comprises:

computing the sensitivity score for each sensitive term; and

upon determining the sensitivity score for the respective sensitive term exceeds a threshold, restricting the plurality of subqueries from including the respective sensitive term.

6. The method of claim 5 , wherein the sensitivity scores are computed based on: (i) the IDF score of the respective sensitive term in a corpus, wherein the sensitivity scores are increased upon determining the IDF score of the respective sensitive term exceeds the predefined frequency threshold, wherein the frequency threshold specifies a threshold number of occurrences of the respective sensitive term in the corpus, and (ii) whether the respective sensitive term is a low level node of the ontology.

7. The method of claim 6 , wherein the first subquery and the second subquery, of the plurality of subqueries, are constructed to offset broadness in the first subquery with specificity in the second subquery by: (i) including, in the first subquery, a hypernym corresponding to a third sensitive term of the plurality of sensitive terms, and (ii) including the third sensitive term in the second subquery.

8. The method of claim 7 , further comprising:

obfuscating the received query by executing the first subquery at a first time and executing the second subquery at a second time, different than the first time.

9. The method of claim 8 , wherein the filtering is performed by a respective server hosting a respective search engine executing the plurality of subqueries.