Abstract: Techniques and mechanisms described herein facilitate the performance metric based placement of bids. According to various embodiments, a first threshold value for bid placement may be identified. The first threshold value may designate a first threshold performance metric level for advertising opportunity bid requests received during a designated time period. An estimated performance metric may be determined for a designated advertising opportunity bid request received during the designated time period. A bid placement message may be transmitted via a communications interface when the estimated performance metric is greater than the first threshold value. The bid request may identify the designated advertising opportunity bid request and a bid placement value.

Abstract: Techniques and mechanisms described herein facilitate efficient budget pacing. According to various embodiments, a plurality of advertising request opportunities may be received at a communications interface during a time interval divided into a plurality of time periods. The time periods may include a designated time period, and a pacing rate for the designated time period may be determined. Each of the advertising request opportunities may be capable of being purchased to present a respective online advertisement. The pacing rate may designate a portion of the advertising request opportunities received during the designated time period on which to bid. One or more instructions may be transmitted to place a plurality of bids on selected ones of the advertising request opportunities received during the designated time period in accordance with the pacing rate.

Abstract: Techniques and mechanisms described herein facilitate adaptive fuzzy fallback stratified sampling. According to various embodiments, an actual or estimated minimum vertex cover of a feature dependency graph representing a dataset may be determined. The dataset may include a plurality of feature vectors and a plurality of features. Each feature vector may include a plurality of feature values that correspond with the features. The feature dependency graph may represent a plurality of conditional dependencies between the features. The minimum vertex cover may designate a subset of the features for strata selection. The feature vectors may be partitioned into a plurality of strata based on the designated subset of features. Each stratum may include one or more of the feature vectors. Each feature vector may be assigned to a corresponding stratum based on the values of the designated subset of features for the feature vector.

Abstract: This disclosure describes systems and methods for identifying and correcting anomalies in web graphs. A web graph is transformed into a set of weighted features. The set of weighted features are then transformed into a signature via a SimHash algorithm. The signature is compared to the signature of one or more other web graphs in order to determine similarity between web graphs. Actions are then carried out to remove anomalous web graphs and modify parameters governing web mapping in order to decrease the likelihood of future anomalous web graphs being built.

Abstract: A bid determination platform receives specifications for advertisements, each associated with a model configured to adjust a parameter value of a bid based on different combinations of user and/or media content characteristics associated with different advertisement requests. The platform also receives an advertisement request for placement of an advertisement on an advertisement space of a publisher's media content that has been requested by a user, and the advertisement request is associated with a set of user and media content profile data. For each of the advertisements, all or a subset of the set of user and/or media content profile data is input to each model so as to determine an associated parameter value. Bids for each advertisement are determined, with at least some bids being based on the associated parameter value. An optimum bid is sent with a location of its advertisement to the advertisement request sender.

Abstract: A data management apparatus for digital advertising includes a data integration processor for collecting and storing data from providers, resolving heterogeneity of the data at schema and data levels, and performing validity checks of the data; an analytics processor for receiving validated data from the data integration processor and providing to users custom, nesting-aware, SQL-like query language and a library of data mining methods, machine learning models, and analytical user profiles (AUP); and an activation processor for encapsulating complex computations performed in real-time, segment evaluation, and online user classification using runtime user profiles (RUP).

Abstract: Techniques for identifying prohibited information within an image are described. For example, a machine accesses an image that depicts an item. The image may include prohibited information that is disallowed in accordance with a policy. The machine identifies the prohibited information within the image based on an analysis of the image. The machine initiates a response to the prohibited information based on the identifying of the prohibited information within the image.

Abstract: Image-based features may be significantly correlated with click-through rates of images that depict a product, which may provide a more formal basis for the informal notion that good quality images will result in better click-through rates, as compared to poor quality images. Accordingly, an image assessment machine is configured to analyze image-based features to improve click-through rates for shopping search applications (e.g., a product search engine). Moreover, the image assessment machine may rank search results based on image quality factors and may notify sellers about low quality images. This may have the effect of improving the brand value for an online shopping website and accordingly have a positive long-term impact on the online shopping website.

Abstract: A metadata extraction machine accesses an image that depicts an item and identifies an attribute of the item and a descriptor that specifies a value for the identified attribute. The machine identifies the attribute and the descriptor based on an analysis of the accessed image that includes distinguishing a foreground portion of the image from a background portion of the image. Distinguishing a foreground portion from the background portion may include segmenting the image to identify the foreground and background portions. Some or all of the analysis may be performed by the machine. The machine then stores the identified descriptor of the item as metadata of the item.

Abstract: The subject matter disclosed herein relates to a system and/or method for allocating data among reduce processes. In a particular implementation, a map process may be executed to provide intermediate data associating key/value pairs with input data. Intermediate data may be allocated among a plurality of reduce processes. At least a portion of intermediate data initially allocated to one or more of said reduce processes may be re-allocated based, at least in part, on a load factor associated with one or more reduce processes.

Abstract: Image-based features may be significantly correlated with click-through rates of images that depict a product, which may provide a more formal basis for the informal notion that good quality images will result in better click-through rates, as compared to poor quality images. Accordingly, an image assessment machine is configured to analyze image-based features to improve click-through rates for shopping search applications (e.g., a product search engine). Moreover, the image assessment machine may rank search results based on image quality factors and may notify sellers about low quality images. This may have the effect of improving the brand value for an online shopping website and accordingly have a positive long-term impact on the online shopping website.

Abstract: Embodiments of the invention present an approach to conversion rate estimation which relies on using past performance observations along user, publisher, and advertiser data hierarchies. More specifically, embodiments of the invention model the conversion event at different select hierarchical levels with separate binomial distributions and estimate the distribution parameters individually. It is shown how to combine these individual estimators using logistic regression to identify conversion events accurately. Embodiments of the invention can also handle many practical issues, such as data imbalance, missing data, and output probability calibration, which render this estimation problem more difficult for a real-world implementation of the approach.

Abstract: The present invention provides a method and system for optimizing search result rankings through use of a game interface. The method and system includes providing a game interface to at least two users, the game interface comprising at least one search query and at least two search result sets. The method and system further includes detecting the selection of one of the two search result sets by the users based on competition criteria and updating ranking data in response to the selection of one of the two search results. The method and system further includes selecting ranking data associated with a given query, determining an optimum ranking based on aggregating the selected ranking data, and storing the optimum ranking.

Abstract: This disclosure describes systems and methods for identifying and correcting anomalies in web graphs. A web graph is transformed into a sequence of tokens via a walk algorithm. The sequence is fingerprinted to form a set of shingles. The singles are compared to shingles for other web graphs in order to determine similarity between web graphs. Actions are then carried out to remove anomalous web graphs and modify parameters governing web mapping in order to decrease the likelihood of future anomalous web graphs being built.

Abstract: A metadata extraction machine accesses an image that depicts an item and identifies an attribute of the item and a descriptor that specifies a value for the identified attribute. The machine identifies the attribute and the descriptor based on an analysis of the accessed image that includes distinguishing a foreground portion of the image from a background portion of the image. Distinguishing a foreground portion from the background portion may include segmenting the image to identify the foreground and background portions. Some or all of the analysis may be performed by the machine. The machine then stores the identified descriptor of the item as metadata of the item.

Abstract: The present invention provides a method and system for optimizing search result rankings through use of a game interface. The method and system includes providing a game interface to at least two users, the game interface comprising at least one search query and at least two search result sets. The method and system further includes detecting the selection of one of the two search result sets by the users based on competition criteria and updating ranking data in response to the selection of one of the two search results.

Abstract: An input data set is treated as a plurality of grouped sets of key/value pairs, which enhances the utility of the MapReduce programming methodology. By utilizing such a grouping, map processing can be carried out independently on two or more related but possibly heterogeneous datasets (e.g., related by being characterized by a common primary key). The intermediate results of the map processing (key/value pairs) for a particular key can be processed together in a single reduce function by applying a different iterator to intermediate values for each group. Different iterators can be arranged inside reduce functions in ways however desired.

Abstract: Image-based features may be significantly correlated with click-through rates of images that depict a product, which may provide a more formal basis for the informal notion that good quality images will result in better click-through rates, as compared to poor quality images. Accordingly, an image assessment machine is configured to analyze image-based features to improve click-through rates for shopping search applications (e.g., a product search engine). Moreover, the image assessment machine may rank search results based on image quality factors and may notify sellers about low quality images. This may have the effect of improving the brand value for an online shopping website and accordingly have a positive long-term impact on the online shopping website.

Abstract: The present invention provides a method and system for optimizing search result rankings through use of a game interface. The method and system includes providing a game interface to at least two users, the game interface comprising at least one search query and at least two search result sets. The method and system further includes detecting the selection of one of the two search result sets by the users based on competition criteria and updating ranking data in response to the selection of one of the two search results. The method and system further includes selecting ranking data associated with a given query, determining an optimum ranking based on aggregating the selected ranking data, and storing the optimum ranking.

Abstract: This disclosure describes systems and methods for identifying and correcting anomalies in web graphs. A web graph is transformed into a sequence of tokens via a walk algorithm. The sequence is fingerprinted to form a set of shingles. The singles are compared to shingles for other web graphs in order to determine similarity between web graphs. Actions are then carried out to remove anomalous web graphs and modify parameters governing web mapping in order to decrease the likelihood of future anomalous web graphs being built.