Back to results
US 2013/0173571 A1Published application
Click noise characterization model
AssigneeMicrosoft Corporation
InventorsWeizhu Chen, Zheng Chen, Adish Singla
Life of this patent
Estimating the term…
Abstract
The techniques discussed herein consider a degree of noise associated with user clicks performed during search sessions. The techniques then generate a model that characterizes click noise so that search engines can more accurately infer document relevance.
Claims20
A method comprising:
- under control of a processor configured with computer-executable instructions,
- receiving, for each of a plurality of query-document pairs identified from one or more search sessions, one or more labels indicating that the query-document pair is relevant or irrelevant;
- extracting feature data for each labeled query-document pair;
- generating a model that characterizes a degree of noise associated with a click event based at least in part on the one or more labels and the feature data extracted for each labeled query-document pair;
- and
- using the model to infer relevance for a plurality of documents associated with a search query.
3 more elements
One or more computer-readable storage media comprising computer-executable instructions configured to perform operations comprising:
- building a noise-aware click model based at least in part on:
- (i) human provided relevance ratings for each of a plurality of query-document pairs in a first group of one or more search sessions, and (ii) a first set of features associated with the first group of one or more search sessions;
- employing the noise-aware click model to characterize a degree of noise for each of a plurality of user click events based at least in part on a second set of features extracted from a second group of one or more search sessions;
- and
- ranking a plurality of documents based at least in part on the respective degrees of noise.
2 more elements
A system comprising:
- one or more processors;
- one or more computer memories, coupled to the one or more processors and storing:
- a feature extraction module, operable by the one or more processors, to extract data corresponding to multiple features from one or more labeled query-document pairs;
- a feature value determination module, operable by the one or more processors, to receive the data from the feature extraction module and determine values for each of the multiple features;
- a noise-aware click model building module, operable by the one or more processors, to receive the values from the feature value determination module and learn noise predictor parameters based at least in part on the one or more labeled query-document pairs and the values;
- and
- a search engine module, operable by the one or more processors, to use the learned noise predictor parameters to infer document relevance for a plurality of search sessions maintained in click-through data logs.
4 more elements
Dependencies are inferred from the claim text.