Prediction of query difficulty for a generic search engine
AssigneeInternational Business Machines Corporation
InventorsDavid Carmel, Lawrence Adam Darlow, Shai Fine, Elad Yom-Tov
Life of this patent
Estimating the term…
Abstract
A query difficulty prediction unit includes a query difficulty predictor to determine the extent of overlap between query documents received from a search engine operating on an input query and sub-query documents received from the search engine operating on sub-queries of the input query. The unit generates a query difficulty prediction from the extent of overlap.
Claims30
A query difficulty prediction unit comprising:
- a processor;
- a memory;
- a query divider to divide an input query into a multiplicity of sub-queries and to provide said input query and said multiplicity of sub-queries to a search engine;
- and
- a query difficulty predictor to generate a query difficulty prediction at least using query documents received from said search engine operating on said input query and said sub-queries, wherein said predictor comprises an overlap counter to generate an overlap vector of the extent of overlap between said query documents for said input query and said query documents for said sub-queries;
- a binary histogram generator to generate a binary histogram of said overlap vector;
- and
- a linear predictor to multiply a per search engine, tuning vector with a ranked version of said binary histogram to generate said query difficulty prediction.
A query difficulty prediction unit comprising:
- a processor;
- a memory;
- and
- a query difficulty predictor to determine the extent of overlap between query documents received from a search engine operating on an input query and sub-query documents received from said search engine operating on sub-queries of said input query and to generate a query difficulty prediction at least from said extent of overlap;
- wherein said query difficulty predictor comprises:
- a binary histogram generator to generate a binary histogram of an overlap vector of said extent of overlap;
- and
- a linear predictor to multiply a per search engine, tuning vector with a ranked version of said binary histogram to generate said query difficulty prediction.
A computer program product readable by a machine, storing a program of instructions on a computer storage medium executable by the machine to perform method steps for query difficulty prediction, said method steps comprising:
- dividing an input query into a multiplicity of sub-queries;
- providing said input query and said multiplicity of sub-queries to a search engine;
- and
- generating a query difficulty prediction at least using query documents received from said search engine operating on said input query and said sub-queries;
- wherein said generating comprises:
- generating an overlap vector of the extent of overlap between said query documents for said input query and said query documents for said sub-queries;
- generating a binary histogram of said overlap vector;
- and multiplying a per search engine, tuning vector with a ranked version of said binary histogram to generate said query difficulty prediction.
A computer program product readable by a machine, storing a program of instructions on a computer storage medium executable by the machine to perform method steps for query difficulty prediction, said method steps comprising:
- determining the extent of overlap between query documents received from a search engine operating on an input query and sub-query documents received from said search engine operating on sub-queries of said input query;
- and generating a query difficulty prediction at least from said extent of overlap;
- wherein said generating comprises:
- generating a binary histogram of an overlap vector of said extent of overlap;
- and
- multiplying a per search engine, tuning vector with a ranked version of said binary histogram to generate said query difficulty prediction.
Dependencies are inferred from the claim text.