Learning-based image webpage index selection
AssigneeMicrosoft Technology Licensing, LLC
InventorsBo Geng, Xian-Sheng Hua, Zhong Wu, Dengyong Zhou
Life of this patent
Estimating the term…
Abstract
Architecture that performs image page index selection. A learning-based framework learns a statistical model based on the hyperlink (URL-uniform resource locator) previous click information obtained from the image search users. The learned model can combine the features of a newly discovered URL to predict the possibility of the newly-discovered URL being clicked in the future image search. In addition to existing web index selection features, image clicks are added as features, and the image clicks are aggregated over different URL segments, as well as the site modeling pattern trees to reduce the sparse problem of the image click information.
Claims20
A system, comprising:
- a data sampling and labeling component configured to sample training webpages having images to create positive webpage samples and negative webpage samples based on user image clicks;
- a feature generation component configured to employ the user image clicks to extract features from the training webpages;
- a model creation component configured to create a selection model for an image webpage index based on the sampled training webpages and the features, wherein the data sampling and labeling component is configured to sample the training webpages within a specified time window, and past user image clicks are used as features to train the selection model;
- a validation component configured to evaluate and validate the selection model for the image webpage index using benchmark data to obtain an optimal selection model;
- a prediction component configured to predict an image webpage static rank of newly-discovered image webpages based on the optimal selection model;
- and
- a microprocessor that executes computer-executable instructions associated with the data sampling and labeling component, feature generation component, model creation component, validation component, and prediction component.
A method, performed by a computer system executing machine-readable instructions, the method comprising acts of:
- sampling and labeling webpages having images as training webpages to create positive webpages and negative webpages based on a user click of an image, the sampling performed within a specified time window;
- employing the user image clicks to extract features from the training webpages;
- creating a selection model for an image index based on the training webpages and the features;
- training the selection model using at least past user image clicks as features;
- evaluating and validating the selection model using benchmark data to obtain an optimal selection model;
- applying the optimal selection model to predict an image webpage static rank of newly-discovered image webpages, such that top ranked newly-discovered image webpages can be selected as targets for the image index based on the image webpage static rank;
- and
- configuring at least one microprocessor to execute instructions to perform the acts of sampling, employing, creating, training, evaluating, and applying.
A method, performed by a computer system executing machine-readable instructions, the method comprising acts of:
- sampling and labeling training webpages having images to create positive webpages and negative webpages derived from user image clicks as labeling criteria the sampling performed within a specified time window;
- extracting historical features from the user image clicks on the training webpages, the historical features of each training webpage accumulated and merged to a specific time;
- creating a statistical selection model based on the sampled training webpages and the features;
- training the selection model using at least past user image clicks as features;
- evaluating and validating the selection model using benchmark data to obtain an optimal selection model;
- applying the optimal selection model to newly-discovered image webpages to compute corresponding selection scores of the newly-discovered image webpages;
- and
- selecting a set of top-ranked newly-discovered image webpages as candidates for indexing based on the scores.
Dependencies are inferred from the claim text.