Abstract:
An example process includes determining a first quality metric that is indicative of a quality of an opportunity for distribution of content from a content provider as compared to other content providers, where the first quality metric is based on a first predicted access rate and a second predicted access rate, where the first predicted access rate is based on features that are dependent on the content provider, and where the second predicted access rate is based on features that are independent of the content provider. The example process also includes determining a second quality metric that is based on the first predicted access rate of the content; determining, a weight to apply to the first quality metric and to the second quality metric; and determining a weighted average of the first quality metric and the second quality metric that is based on the weight.
Abstract:
Methods, systems, and apparatus include computer programs encoded on a computer-readable storage medium, including a method for determining bid prices for content items. A winner associated with an auction is identified for delivery of a content item, including identifying a first bid associated with the winner and an associated first expected clickthrough rate. A next finisher is identified in the auction including identifying a second bid associated with the next finisher and a second expected clickthrough rate associated with second bid. A price is determined that a content sponsor associated with the winner should pay for presentation of the content item, including identifying a third bid and corresponding third expected clickthrough rate dependent on the third bid whose product, being the first product, is substantially equal to a product, being the second product, of the second bid and the second expected clickthrough rate. The content sponsor is charged the price.
Abstract:
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for providing a content item. In one aspect, a method includes receiving a content item request. A set of candidate content items that are eligible to be provided in response to the content item request is identified. A performance measure is predicted for each candidate content item based at least in part on a loss function that specifies an economic cost of incorrectly predicting the performance measure for the candidate content item. The loss function can be based in part on a distribution of competing bid values for a set of previous content item impressions. A candidate content item can be selected for presentation based on the predicted performance measure for the candidate content items. The selected candidate content item is provided in response to the content item request.
Abstract:
Systems and methods of optimizing a configuration of content items for display with an online document are provided. A system can identify from a content database a configuration for the online document and a configuration attribute for the configuration, which can include one or more content item slots. The system can determine a selection factor for the configuration based on the configuration attribute. The system can obtain a bid value for a content item and a click-through attribute for the content item slot of the content item configuration. The system can select the configuration based on the selection factor, the bid values for the content item, and the click-through attributes for the content item slot. The system can authorize the content item slot of the configuration to include a content item. The system can provide the online document with the content item for display by a computing device.
Abstract:
A method for setting allocations and prices for an auction including receiving a request for content for presentation in association with one or more presentation opportunities on a publisher site, determining one or more eligible content items based on the received request, determining bids associated with each eligible content item, determining a historical bid distribution for bids that have been identified when selecting content for presentation on the publisher site, determining a risk adjustment parameter that reflects a measure of confidence in the accuracy of the historical bid distribution, conducting an auction including scoring the determined eligible content items based on the historical bid distribution, their respective bids, and the risk adjustment parameter, selecting a winning content item from the eligible content items based on the auction, setting a price for the winning content item, and providing the winning content item in response to the request.
Abstract:
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for distributing content are disclosed. In one aspect, a method includes receiving a content item request specifying that at least two presentation positions are available for presentation of content items. Current bids specifying amounts that two or more content sponsors are willing to pay to provide a content item in response to the content item request are identified. For each of the two or more content sponsors, a sponsor value is determined based, at least in part, on the current bid received for the content sponsor and one or more previous bids that were previously received for the content sponsor. At least one content sponsor is selected to provide a content item based at least in part on the sponsor values.
Abstract:
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for ranking content items. In one aspect, a method includes identifying, for a content item, a bid value specifying an amount a content item provider is willing to pay for user interaction with the content item. A predicted performance measure is identified for the content item. The predicted performance measure is adjusted based on a weighting factor for the content item. The weighting factor for the content item is indicative of confidence that the predicted performance measure will match an actual performance measure for the content item and can be different than a weighting factor for another content item identified for inclusion in a ranking with the content item. A rank score is determined for the content item using the bid value and adjusted predicted performance measure. The content item is provided based on the rank score.
Abstract:
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for distributing content through an exchange are disclosed. In one aspect, a method includes submission of a first bid to a content item selection process and receiving a request for one or more content items based on the first bid. A minimum price is determined based on a second bid that was submitted to the content item selection process by another entity. A set of content items having bids that meet the minimum price are identified, and a configuration of one or more content items that provides a threshold efficiency is identified. For each particular content item in the configuration of content items, a price that will be paid for distribution of the particular content item in the configuration is determined, and the price for at least one particular content item meets the minimum price.
Abstract:
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for distributing content items are disclosed. In one aspect, a method includes accessing a scaling factor value and accessing a first page value range specifying at least a high page value and a low page value. A determination is made that a first ranking of content items based on the high page value does not match a second ranking of the content items that is based on the low page value. In response to determining that the first ranking does not match the second ranking, an updated first ranking and an updated second ranking are determined based on a second page value range. A determination is made that the updated first ranking matches the updated second ranking. Content items are distributed based on the updated first ranking.
Abstract:
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for optimizing machine learning systems. In one aspect a method includes determining an average error of a machine learning system (“MLS”). An evaluation function that provides a result that would have been achieved using a specified value of a given parameter is defined. An expected outcome function that provides expected results for prior events based on the error of the MLS is defined. For each of multiple prior events, a target value of the given parameter is determined, e.g., using the expected outcome function. A model is generated using the MLS based on features of the prior events and the determined target values of the given parameter for the prior events. A value is assigned to the given parameter for a new event based on application of the model to features of the new event.