Loading patent…
AssigneeMicrosoft Corporation
InventorsYutaka Suzue, Johnson T. Apacible, Mark J. Encarnación, Jamie Huynh, Simon D. Bernstein
Estimating the term…
The subject disclosure is directed towards a technology by which the accuracy of context-based information provided by at least one data source for received context data is increased. Correctness information received in association with usage of looked up context-based information is logged. The correctness information may be processed to increase the overall accuracy by correcting a data source, and/or by creating a blended data source that includes the most likely accurate portions (segments) from among multiple data sources as determined via the correctness information.
In a computing environment, a method performed at least in part on at least one processor, comprising, receiving context data that correspond to context-based information provided by at least one data source, receiving correctness information corresponding to accuracy of the context-based information, and processing the correctness information to increase an overall accuracy corresponding to other context-based information provided by the at least one data source for other received context signals.
A system comprising, a sampling service, the sampling service configured to log correctness data for at least some of a plurality of sample requests, in which each sample request is associated with context data, and for each sample request for which information is logged, the sampling service configured to obtain, from at least one entity other than the one or more data sources, a measure of correctness related to context-based information looked up in one or more data sources based upon the context data associated with that request, and to log data representing the measure of correctness.
One or more computer-readable media having computer-executable instructions, which when executed perform steps, comprising, creating a blended data source from among a plurality of data sources that each contain at least some inaccurate data, including segmenting each data source into segments to provide a plurality of counterpart segments among the data sources, processing correctness data collected relative to each counterpart segment to determine a measure of correctness for each counterpart segment, and selecting selected segments for the blended data source from among the counterpart data sources based upon the measure of correctness for each counterpart segment.
Dependencies are inferred from the claim text.