JP2018041509A - Automated click type selection for content performance optimization - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide systems and methods for optimizing content performance.SOLUTION: A content generation system receives request for a content item to be presented on a client device. The system uses a click type performance model and the serving context for the content item to calculate a predicted performance metric for the content item for multiple different potential click types. The system selects one of the potential click types on the basis of the predicted performance metrics and generates a content item having the selected click type. The click type performance model is updated using event data from the client device.SELECTED DRAWING: Figure 1

Description

  In computerized content distribution networks, content providers typically design content items (eg, advertisements) and provide them to content servers. The content server then distributes the content item to various client devices through one or more content slots of electronic resources. Often, content items include images, videos, graphics, text, and / or other visual representations. It can be difficult and difficult for content providers to generate effective and attractive content items.

  One implementation of the present disclosure is a method for optimizing content performance using click types. The method includes receiving at a processing circuit a request for a content item for presentation on a client device. The request includes an indication of a serving context for the content item. The method uses a click type performance model for a content item and a service provisioning context to calculate, by a processing circuit, a measure of the expected performance of the content item for a plurality of different potential click types. The method further includes the step of using Each potential click type defines different criteria for triggering an event that occurs in response to a user interaction with the content item when the defined criteria are met. The method further includes selecting one of the potential click types by a processing circuit based on an expected performance metric and generating a content item having the selected click type. The content item is generated by the processing circuit in response to the request. The method further includes providing the generated content item for presentation on the client device.

  In some implementations, the method collects and stores event data indicating user actions associated with content items having each of the potential click types, and trains the event data click type performance model. Using as feedback for.

  In some implementations, the different criteria defined by the potential click type include different types of user interaction with the content item. In some implementations, each potential click type defines a different criterion for calculating a performance metric.

  In some implementations, selecting one of the potential click types includes selecting a button click type. When the button click type is selected, the method treats the event as a user interaction with the first part of the content item excluding the user's interaction with a second part of the content item that is separate from the first part. It may further include defining criteria for triggering and rendering a clickable button that occupies only the first portion of the content item.

  In some implementations, selecting one of the potential click types includes selecting an entire content item click type. When the entire content item click type is selected, the method includes the steps of defining criteria for triggering an event as a user interaction with any part of the content item, and a clickable element throughout the display area of the content item. Rendering as:

  In some implementations, selecting one of the potential click types includes selecting a confirm click type. When a confirmation click type is selected, the method displays a confirmation prompt in response to the first user interaction with the content item, and a confirmation prompt is displayed after the first user interaction. Defining criteria for triggering the event as a second user interaction with the content item in between.

  In some implementations, the service provisioning context includes at least one of the type of client device, the platform of the client device, and the resource with which the content item is presented in association.

  In some implementations, selecting one of the potential click types includes selecting a click type associated with an extreme of the expected performance metric for the potential click type. .

  In some implementations, the method further includes comparing the predicted performance metric to a threshold. Selecting one of the potential click types includes selecting a first click type of potential click types if the predicted performance metric is below a threshold, and predicting Selecting a second click type of potential click types when the performance metric to be performed is greater than or equal to a threshold value.

  Another implementation of the present disclosure is a system for optimizing content performance using click types. The system includes a processing circuit having a processor and a memory. The processing circuit is configured to receive a request for a content item for presentation on a client device. The request includes an indication of the service provision context for the content item. The processing circuitry is further configured to use the click type performance model and the service delivery context for the content item to calculate the expected performance metric of the content item for a plurality of different potential click types. . Each potential click type defines different criteria for triggering an event that occurs in response to a user interaction with the content item when the defined criteria are met. The processing circuit selects one of the potential click types based on the expected performance metric and generates a content item with the selected click type in response to the request on the client device It is further configured to provide a content item generated for presentation.

  In some implementations, the processing circuitry collects and stores event data indicating user actions associated with content items having each of the potential click types, and trains the event data click type performance model. Configured to be used as feedback.

  In some implementations, the different criteria defined by the potential click type include different types of user interaction with the content item. In some implementations, each potential click type defines a different criterion for calculating a performance metric.

  In some implementations, selecting one of the potential click types includes selecting a button click type. When the button click type is selected, the processing circuit events as a user interaction with the first part of the content item excluding the user's interaction with the second part of the content item that is separate from the first part. Can be configured to render a clickable button that occupies only the first portion of the content item.

  In some implementations, selecting one of the potential click types includes selecting an entire content item click type. When the entire content item click type is selected, the processing circuitry defines criteria for triggering an event as a user interaction with any part of the content item, and as a clickable element for the entire display area of the content item. Can be configured to render.

  In some implementations, selecting one of the potential click types includes selecting a confirm click type. When the confirmation click type is selected, the processing circuitry configures the content item to display a confirmation prompt in response to the first user interaction with the content item, and after the first user interaction, the confirmation It may be configured to define criteria for triggering an event as a second user interaction with the content item that occurs while the prompt is displayed.

  In some implementations, the service provisioning context includes at least one of the type of client device, the platform of the client device, and the resource with which the content item is presented in association.

  In some implementations, selecting one of the potential click types includes a click type associated with an extreme value of the predicted performance metric for the potential click type.

  In some implementations, the processing circuitry is configured to compare the predicted performance metric to a threshold. Selecting one of the potential click types includes selecting the first click type of potential click types if the predicted performance metric is below a threshold, and predicting Selecting a second click type of potential click types when the performance metric to be performed is greater than or equal to a threshold value.

  Another implementation of the present disclosure is a method for optimizing content performance using creative extensions. The method includes receiving at a processing circuit a request for a content item for presentation on a client device. The request includes an indication of the service provision context for the content item. The method includes a processing circuit that calculates a creative extension performance model for the content item and a service provisioning context to calculate a measure of the expected performance of the content item for a plurality of different potential creative extensions. The method further includes the step of using Each potential creative extension defines a different action to be performed in response to a user interaction with the content item. The method further includes selecting one of the potential creative extensions by a processing circuit based on a predicted performance metric, and generating a content item having the selected creative extension. The content item is generated by the processing circuit in response to the request. The method further includes providing the generated content item for presentation on the client device.

  In some implementations, the method collects and stores event data indicating user actions associated with content items having each of the potential creative extensions, and trains the event data into the creative extension performance model. Using as feedback for.

  In some implementations, the method includes extracting a data asset from at least one of a landing resource associated with the content item, a secondary resource associated with the landing resource, and a customer information database. The method may include using the extracted data asset to determine which of the potential creative extensions can be utilized for selection. In some implementations, using the extracted data assets to determine which of the potential creative extensions can be utilized for selection is required to generate the creative extension Responding to the steps of identifying the data asset, determining whether the identified data asset has been extracted, and determining that the required data asset has not been extracted, the creative extension is used for selection Determining that it cannot be performed.

  In some implementations, selecting one of the potential creative extensions can use any of the potential creative extensions for selection using the service-providing context for the content item. The step of determining.

  In some implementations, selecting one of the potential creative extensions includes selecting a click-to-call extension. When a click-to-call extension is selected, the method uses the extracted contact information to generate a creative extension configured to initiate communication with a subject associated with the content item; Rendering the rendered creative extension as a selectable element of the content item.

  In some implementations, selecting one of the potential creative extensions includes selecting a position extension. When a location extension is selected, the method uses the extracted location information to launch a map application or generate a creative extension that is configured to display the geographic location associated with the content item. And rendering the generated creative extension as a selectable element of the content item.

  In some implementations, selecting one of the potential creative extensions includes selecting a secondary resource extension. When a secondary resource extension is selected, the method uses the extracted secondary resource information to generate a creative extension configured to navigate the client device to a secondary resource associated with the content item; Rendering the generated creative extension as a selectable element of the content item.

  In some implementations, selecting one of the potential creative extensions includes selecting a creative extension associated with an extreme of a predicted performance metric for the potential creative extension. .

  In some implementations, the service provisioning context includes at least one of the type of client device, the platform of the client device, and the resource with which the content item is presented in association.

  Another implementation of the present disclosure is a system for optimizing content performance using creative extensions. The system includes a processing circuit having a processor and a memory. The processing circuit is configured to receive a request for a content item for presentation on a client device. The request includes an indication of the service provision context for the content item. The processing circuitry is further configured to use the creative extension performance model and the service provision context for the content item to calculate a measure of the expected performance of the content item for a plurality of different potential creative extensions. . Each potential creative extension defines a different action to be performed in response to a user interaction with the content item. The processing circuit selects one of the potential creative extensions based on the expected performance metric and generates a content item with the selected creative extension in response to the request on the client device. It is further configured to provide a content item generated for presentation.

  In some implementations, the processing circuitry collects and stores event data indicating user actions associated with content items having each of the potential creative extensions, and trains the event data into the creative extension performance model. Configured to be used as feedback.

  In some implementations, the processing circuitry is configured to extract data assets from at least one of a landing resource associated with the content item, a secondary resource associated with the landing resource, and a customer information database. The processing circuit may use the extracted data asset to determine which of the potential creative extensions can be utilized for selection. In some implementations, using the extracted data assets to determine which of the potential creative extensions can be utilized for selection is required to generate the creative extension Responsive to identifying the data asset, determining whether the identified data asset has been extracted, and determining that the required data asset has not been extracted, the creative extension is used for selection Determining that it cannot be done.

  In some implementations, selecting one of the potential creative extensions can use any of the potential creative extensions for selection using the service-providing context for the content item. Including determining.

  In some implementations, selecting one of the potential creative extensions includes selecting a click-to-call extension. When the click-to-call extension is selected, the processing circuit uses the extracted contact information to generate and generate a creative extension configured to initiate communication with the subject associated with the content item. Creative extensions may be configured to render as selectable elements of the content item.

  In some implementations, selecting one of the potential creative extensions includes selecting a position extension. When a location extension is selected, the processing circuit uses the extracted location information to launch a map application or a creative extension configured to display the geographic location associated with the content item. Generating and rendering the generated creative extension as a selectable element of the content item.

  In some implementations, selecting one of the potential creative extensions includes selecting a secondary resource extension. When a secondary resource extension is selected, the processing circuit uses the extracted secondary resource information to generate and generate a creative extension that is configured to navigate the client device to the secondary resource associated with the content item. Rendered creative extensions can be configured to render as selectable elements of the content item.

  In some implementations, selecting one of the potential creative extensions includes selecting a creative extension that is related to the extreme of the expected performance metric for the potential creative extension. .

  In some implementations, the service provisioning context includes at least one of the type of client device, the platform of the client device, and the resource with which the content item is presented in association.

  The above is a summary and therefore includes simplifications, generalizations, and omissions of details as appropriate. As a result, those skilled in the art will appreciate that the summary is merely illustrative and not intended to be limiting in any way. Other aspects of the devices and / or processes described herein, as defined solely by the claims, features of the invention, and advantages, are described in this specification and are to be interpreted in conjunction with the accompanying drawings. It will become clear in the explanation.

FIG. 2 is a block diagram of a computer system having a content generation system configured to optimize content performance using click types and creative extensions, in accordance with the described implementation. FIG. 2 is a block diagram illustrating the content generation system of FIG. 1 that trains a predictive model for content performance optimization using feedback from client devices, in accordance with the described implementation. FIG. 2 is a block diagram illustrating several components of the content generation system of FIG. 1 including a content extractor, creative selector, click type selector, and creative extension selector, in accordance with the described implementation. FIG. 4 is a block diagram illustrating the content extractor of FIG. 3 in more detail according to the described implementation. FIG. 4 is a block diagram illustrating a creative generator that may be used by the creative selector of FIG. 3 to automatically generate creatives using extracted data assets, in accordance with the described implementation. FIG. 4 is a block diagram illustrating the click type selector of FIG. 3 in more detail according to the described implementation. FIG. 4 is a block diagram illustrating in more detail the creative extension selector of FIG. 3 in accordance with the described implementation. 4 is a flow diagram of a process for optimizing content performance using click types, according to a described implementation. 4 is a flow diagram of a process for optimizing content performance using creative extensions in accordance with the described implementation.

  Referring generally to the figures, systems and methods for automated content generation and optimization in accordance with various implementations are shown. The systems and methods described herein are specific content providers, specific resources (eg, web pages), specific platforms (eg, display content, video content, mobile device content, text content, etc.), specific It can be used to automatically generate a content item (eg, an advertisement) tailored to a profile of interest (eg, a set of user interests), a particular impression, or any combination thereof. A content item may be generated by a content generation system, provided to a client device, and presented to a user via the client device. The generated content item is responsive to the user's interaction with the landing resource (for example, clicking on the content item, hovering over the content item, etc.) (for example, by an embedded hyperlink) on the client device To a landing resource.

  In some implementations, the content generation system extracts data assets from landing resources and / or other related data sources (eg, secondary resources associated with landing resources, separate databases, etc.). The resource from which the data asset is extracted may be defined by the identification information of the landing resource. Extracted data assets include images, colors, fonts, text, graphics, videos, styles, company related information (e.g., geographical location, contact information, etc.), page classification, and / or landing resources Other types of related data may be included. The content generation system uses the extracted data assets to direct landing resources and / or related secondary resources in response to user interaction with the generated content items (eg, via embedded hyperlinks) Content items can be generated.

  User actions related to generated content items (for example, clicks, conversions, and other types of events) are logged to optimize the performance of subsequent content items generated by the content generation system Can be used. The performance of the generated content item may be quantified using various performance metrics and / or statistical data derived from logged user actions. One example of such a performance metric is the click rate (CTR) associated with a content item, a group of content items, or a content distribution campaign. The CTR for a content item is the percentage of content item clicks that lead to a conversion (i.e.

). A conversion can be any event or user action that meets predetermined conversion criteria (for example, purchasing a product, creating a user account, threshold time spent looking at a particular resource, etc.) There is sex.

  In some implementations, content providers are charged for content items based on cost-per-click (CPC). But not every click leads to a conversion. In a CPC billing system, the CTR for a content item may be correlated with the return on investment (ROI) realized by the content provider. In general, content providers seek to maximize the ROI associated with their content provider's content distribution campaigns. One way to maximize ROI is to optimize content selection and delivery decisions with the goal of maximizing various performance metrics such as CTR.

  The content generation system described herein may optimize content performance by automatically adjusting various elements, features, or attributes of generated content items. In some implementations, the content generation system uses one or more prediction models to calculate an estimated performance metric for a particular feature or combination of features of the generated content item. The content generation system may use the predictive model to select the best features to include in the generated content item for a particular service delivery context. Predictive models include landing resource ID (e.g. URL) or type (e.g. page classification), device type (e.g. mobile, tablet, desktop etc.), content platform (e.g. video, text, image etc.), device location May be configured to receive input, such as user interest, or other information indicating service delivery context and / or available impressions. The predictive model may be used to calculate a predicted performance metric (eg, predicted click rate (pCTR)) for various content item features based on a given service delivery context.

  In some implementations, the content generation system trains the predictive model using logged events and / or statistical data associated with previously generated content items. The logged event data can be used as a feedback signal for the predictive model and used to adjust various parameters of the predictive model (e.g., using machine learning techniques) there is a possibility. Training a predictive model using actual event data can improve the accuracy of the prediction and when new events are logged (for example, at regular intervals or continuously) It may allow the prediction model to be updated.

  One feature of a content item that can be optimized by a predictive model is the “click type” of the content item. The click type for a content item can affect the layout of the content item, the interactivity of the content item, and / or how the content item is rendered or presented by the client device. In some implementations, the click type for a content item defines the type of user action that qualifies as a “click” for the purpose of charging the content provider and calculating the pCTR. The content generation system may select a click type from a number of different click types, including a “button” click type, an “entire content item” click type, and a “confirm” click type.

  A button click type may define a click as a user interaction with a particular part of a content item, such as a clickable button or other subset of the entire area of the content item. For content items that have a button click type, user interaction with other parts of the content item (for example, parts outside the clickable area) may not cause the client device to navigate to the landing resource and charge And / or may not qualify as a click for the purpose of calculating performance metrics.

  The entire content item click type may define a click as a user interaction with any part of the content item. For content items that have the entire content item click type, user interaction with any part of the content item may cause the client device to navigate to the landing resource, and calculate billing and / or performance metrics. May be eligible for the intended click.

  The confirmation click type may cause the content item to display a confirmation prompt in response to the first user interaction with the content item (eg, the first click). The confirmation prompt may display a message instructing the user to click again to navigate to the landing resource. For content items that have a confirm click type, a second user interaction with the content item (e.g., a second click) may cause the client device to navigate to a landing resource, and charge and / or performance May be eligible for a click to calculate metrics. Confirmation click types can help protect against unintentional clicks that are unlikely to lead to conversions.

  Another content item feature that can be optimized by the predictive model is the “creative extension” feature of the content item. Creative extensions can be supplemental elements of content items other than the main “creative” part. Creative extensions may include clickable buttons such as “click to call” buttons, “position” buttons, “secondary resources” buttons, and / or other types of supplemental content that may be included in a content item. Each creative extension may cause a different action to be performed when the corresponding part of the content item is clicked.

  A click-to-call button may be included for a content item that is displayed on a mobile communication device when a telephone number or other contact information regarding the subject associated with the content item is known. Such contact information may be automatically extracted from landing resources or obtained from a separate database. Selecting a click-to-call button may initiate a call or other communication with an entity (eg, retail store, customer service phone, sales agent, etc.) associated with the content item. For other device types (eg, desktop computers, tablets, etc.), when the contact information is known, a “click-to-email” button or wider than the click-to-call button A “click-to-contact” button may be included.

  The location button may display a map showing the company's geographic location associated with the landing resource. The geographical location of the associated company (eg, street address, store location, email address, GPS coordinates, etc.) may be extracted from landing resources, separate databases, or resources related to landing resources. Selecting the location button may cause the map application to load (eg, on a mobile device) or may navigate the user's web browser to a map resource (eg, Google map).

  The secondary resource button may display the secondary resource. A content item may include creative extensions (eg, selectable buttons or links) for a number of different resources or resources (eg, secondary resources, social resources, landing resources, etc.). In some implementations, content item links link only to secondary resources (not landing resources) and may be referred to as social content items. In other implementations, content item links link to both landing resources and secondary resources.

  Content items generated by the content generation system are provided to client devices via a communication network (eg, the Internet). Events associated with the content item (eg, impressions, clicks, conversions, etc.) are logged and stored in a database that can be accessed by the content generation system. The content generation system uses the logged events to calculate performance metrics associated with the content item and train the predictive model. A content generation system uses a predictive model to estimate which features or combination of features of a content item will lead to optimal content performance for a given service delivery context (e.g., CTR, pCTR, or Determine the feature that optimizes another performance metric. Then, the content generation system can automatically generate a content item having characteristics estimated to lead to optimal content performance.

  Referring now to FIG. 1, a block diagram of a computer system 100 according to the described implementation is shown. Computer system 100 is shown to include network 102, content requester 104, resource 106, client device 108, resource renderer 110, data storage device 112, and content generation system 114.

  Network 102 may include any type of computer network, such as a local area network (LAN), a wide area network (WAN), a cellular network, a satellite network, a wireless network, the Internet, or any other type of data network. Network 102 may include any number of computing devices (eg, computers, servers, routers, network switches, etc.) configured to transmit, receive, or relay data. Network 102 may further include any number of wired and / or wireless connections. For example, content requester 104 may communicate wirelessly (eg, via WiFi, cellular, radio, etc.) with a transceiver wired to computing devices of network 102 (eg, via fiber optic cable, CAT5 cable, etc.).

  The network 102 may facilitate communication between the content generation system 114 and the content requester 104. The content generation system 114 may receive a content generation request from the content requester 104 via the network 102. Content generation system 114 may generate a content item in response to the request and provide the generated content item to content requester 104 for review or approval.

  The network 102 may also facilitate communication between the content generation system 114, the resource 106, and the resource renderer 110. Content generation system 114 may receive data assets from resource 106 and / or resource renderer 110. Upon receiving the request for content generation, the content generation system 114 may call the resource renderer 110 to obtain (eg, download) and render data (eg, HTML data, web page data, etc.) from the resource 106. Resource renderer 110 may receive data from resource 106 via network 102 and render such data as a snapshot image (e.g., a visual representation of resource 106) and / or as a document object model (DOM) tree. . The rendered data can be transmitted from the resource renderer 110 to the content generation system 114 via the network 102.

  With continued reference to FIG. 1, the computer system 100 is shown to include a content requester 104. Content requester 104 may include one or more entities from which requests for generating content items are received. For example, content requester 104 may include an advertiser, advertising agency, third party content provider, publisher, website provider, or any other entity that may receive a request to generate a content item. There is.

  In some implementations, the content requester 104 includes one or more electronic devices (eg, computers, computer systems, servers, etc.) that can send requests for content generation. Content requester 104 may include a user input device (eg, keyboard, mouse, microphone, touch screen, tablet, smartphone, etc.) from which a user may enter a content generation request. The content requester 104 can send a content generation request to the content generation system 114 via the network 102. In some implementations, the content generation request includes a uniform resource locator (URL). The URL may specify the location of a particular landing resource (eg, one of resources 106).

  In some implementations, the content requester 104 sends campaign parameters to the content generation system 114. Campaign parameters may be used to control the delivery of content items generated by content generation system 114. The campaign parameters may include keywords associated with the content item, bids associated with the keyword, content distribution budget, geographic limiter, or content generation system 114, or when the content item may be presented to the client device 108. May include other criteria used by a separate content server to determine.

  Content requester 104 may access content generation system 114 to monitor the performance of content items delivered according to established campaign parameters. For example, content requester 104 may access content generation system 114 to review one or more performance metrics associated with the content item or set of content items. The performance metric may indicate an interaction (eg, CTR, impressions, clicks, conversions, amount spent, etc.) between client devices 108 associated with the content item or set of content items being delivered. Performance metrics may be based on user actions logged and processed by an accounting system or log file processing system.

  With continued reference to FIG. 1, computer system 100 is shown including resources 106. Resource 106 may include any type of information or data structure that may be provided over network 102. In some implementations, the resource 106 may be identified by a resource address such as a uniform resource locator (URL). Resource 106 can be a web page (e.g., HTML web page, PHP web page, etc.), document processing document, portable document format (PDF) document, image, video, programming element, interactive content, streaming video / audio source, or other May contain types of electronic information.

  Resources 106 may include local resources, intranet resources, Internet resources, or other network resources. In some implementations, the resource 106 is one or more to which the client device 108 is directed (eg, by an embedded hyperlink) when the client device 108 interacts with a content item generated by the content generation system 114. Including web pages. In some implementations, the resource 106 provides additional information about a product, service, or company that is specifically sold within the generated content item. For example, resource 106 may be a web page or website where a product or service specifically sold within the generated content item can be purchased.

  Resources 106 may include landing resources and / or secondary resources. In some implementations, landing resources are specified by the content requester 104 as part of a request to generate a content item. The landing resource may be specified as a URL that leads to one of the resources 106 or otherwise specifies the location of the landing resource. The landing resource URL may be included as part of the content generation request. A secondary resource may be associated with a specified landing resource. Secondary resources may include any type of resource 106 (eg, web page, website, document processing document, video, PDF document, etc.) and may be associated with a primary landing resource. In various implementations, the association between the landing resource and the secondary resource may be a data asset extracted from the landing resource (e.g., a hyperlink that links from the landing resource to the secondary resource) and / or (e.g., the data storage device 112 Can be identified and / or generated using a previously stored association between the landing resource and the secondary resource. In some implementations, the content generation system 114 accesses the data storage device 112 to identify secondary resources associated with the specified landing resource.

  In some implementations, the resource 106 may be combined with the content requester 104. For example, resource 106 may include data stored on one or more electronic devices (eg, computers, servers, etc.) maintained by content requestor 104. In other implementations, the resource 106 may be separated from the content requester 104. For example, resource 106 may be a remote server (eg, FTP server, file sharing server, web server, etc.), a combination of servers (eg, data center, cloud computing platform, etc.), or other data separated from content requester 104. May contain data stored on the storage device.

  With continued reference to FIG. 1, computer system 100 is shown to include client device 108. Client device 108 may include any number and / or type of electronic devices operable by a user. For example, the client device 108 interacts with a desktop computer, laptop computer, smartphone, tablet, mobile communication device, remote workstation, client terminal, entertainment console, or other component of the computer system 100 (eg, via a communication interface). Any other device capable of doing so may be included. Client device 108 may be able to receive resource content from resources 106 and / or third party content items generated by content generation system 114. Client device 108 may include a mobile device or a non-mobile device.

  In some implementations, the client device 108 is an application (e.g., web browser, resource renderer, etc.) for converting electronic content into a user understandable format (e.g., visual, audio, graphical, etc.). )including. Client device 108 can present content to the user, receive user input, or facilitate user interaction with electronic content (eg, clicking on a content item, hovering over a content item, etc.) User interface elements (eg, electronic displays, speakers, keyboards, mice, microphones, printers, etc.). Client device 108 may function as a user agent to allow a user to view content encoded in HTML.

  Client device 108 may include a processor that can process embedded information (eg, meta-information embedded in a hyperlink) and execute embedded instructions. The embedded instructions may include computer readable instructions (eg, software code, JavaScript, ECMAScript, etc.) associated with the content slot in which the third party content item is presented.

  In some implementations, the client device 108 can detect interaction with the delivered content item. The interaction with the content item can be to display the content item, hover over the content item, click the content item, view source information about the content item, or between the client device 108 and the content item. Any other type of interaction may be included. Interaction with a content item does not require explicit action by the user for a particular content item. In some implementations, impressions (eg, displaying or presenting content items) may be eligible for interaction. Criteria for defining which user actions (e.g., active or passive) are eligible for interaction are determined individually (e.g., for each content item) by the content requester 104 or the content generation system 114. There is a possibility.

  The client device 108 may generate event data that indicates a user action regarding the content item. For example, the client device 108 may generate event data in response to the detected interaction with the content item. The event data may include multiple attributes including a content identifier (eg, content ID or signature element), device identifier, reference URL identifier, time stamp, or any other attribute that indicates an interaction. In some implementations, the event data includes an indication of service delivery context (e.g., device type, platform, etc.) and one or more attributes (e.g., click type, creative extension, etc.) of the provided content item. Includes instructions. Client device 108 may generate event data when certain actions (eg, viewing resources, purchasing online, sending a search query, etc.) are performed by the client device. Event data generated by the client device 108 may be communicated to the content generation system 114 or a separate accounting system. The content generation system 114 may use the event data to train a predictive model to optimize content performance.

  With respect to situations where the system discussed herein collects or utilizes personal information about the user, the user may determine that the program or feature is user information (e.g., the user's social network, social action or activity). Whether or not to collect content from the content server that may be more relevant to the user and / or You may be given the opportunity to control how you should receive it. In addition, certain data may be processed in one or more ways (eg, by content generation system 114) before being stored or used so that personally identifiable information is deleted. There is. For example, if the user's identification information cannot determine personally identifiable information about the user, or if the location information is obtained, the user's geographic location (by city, zip code, or state level) Etc.) and may therefore be processed such that the user's specific location cannot be determined. Thus, the user may be able to control how information is collected (eg, by the application, by the client device 108, etc.) and used by the content generation system 114.

  With continued reference to FIG. 1, computer system 100 is shown including a resource renderer 110. Resource renderer 110 may be a hardware or software component that can interpret resource 106 and generate a rendered representation (eg, image, display, etc.) of that resource 106. For example, resource 106 may include marked-up content (eg, HTML, XML, image URL, etc.) and format information (eg, CSS, XSL, etc.). The resource renderer 110 may download the marked up content and format information and render the resource 106 according to the World Wide Web Consortium (W3C) standard. Resource renderer 110 may generate a “snapshot image” of resource 106 and / or build a document object model (DOM) tree that represents resource 106.

  A snapshot image can be a visual representation of a particular landing resource. The snapshot image may show the appearance of the landing resource presented on the user interface device (eg, electronic display screen, computer monitor, touch display, etc.) after rendering the landing resource. A snapshot image may include color information (eg, pixel color, brightness, saturation, etc.) and style information (eg, right angle corners, rounded edges, modern, naive, etc.) regarding landing resources. In some implementations, the snapshot image can be an image file with any viable file extension (eg, .jpg, .png, .bmp, etc.).

  A DOM tree can be a hierarchical model of a particular landing resource. A DOM tree may specify multiple data assets for a landing resource. The DOM tree contains image information about landing resources (e.g., image URL, display position, display size, alternate text, etc.), font information (e.g., font name, size, effects, etc.), color information (e.g., RGB color values, 16 Hex color code), and text information. The DOM tree may include data assets (eg, metadata, hyperlink URLs, object attributes, etc.) that are not directly visible in the snapshot image. The resource renderer 110 may generate a rendered representation of the landing resource that includes all of the landing resource's data assets. Data assets may be provided to the content generation system 114 as features of nodes and / or snapshot images of the DOM tree.

  In various implementations, the resource renderer 110 may be part of the content generation system 114, the client device 108, or a separate component. Resource renderer 110 may prepare a snapshot image and / or DOM tree in response to a rendering request from content generation system 114. Resource renderer 110 may send a snapshot image and / or DOM tree to content generation system 114 in response to the rendering request.

  With continued reference to FIG. 1, computer system 100 is shown including a data storage device 112. Data storage device 112 may store profile data, content item data, accounting data, or any other type of data used by content generation system 114 or another component of computer system 100. It can be any kind of memory device that can. Data storage device 112 may include any type of non-volatile memory, media, or memory device. For example, the data storage device 112 may be a semiconductor memory device (eg, EPROM, EEPROM, flash memory device, etc.), a magnetic disk (eg, internal hard disk, removable disk, etc.), a magneto-optical disk, and / or a CD ROM and DVD. -May include ROM disk.

  In some implementations, the data storage device 112 is local to the content generation system 114, the resource 106, or the content requester 104. In other implementations, the data storage device 112 is a remote data storage device connected to the content generation system 114 and / or the content requester 104 via the network 102. In some implementations, the data storage device 112 is part of a data storage server or system that can receive queries from the content generation system 114 and / or content requestor 104 and respond to the queries.

  In some implementations, the data storage device 112 stores data assets extracted from the resource 106. For example, the data storage device 112 may store image data relating to various images displayed on the resource 106. The image data may include the actual image (eg, an image file), the URL location of the image, image attributes, image metadata, or other attributes of the image displayed on the resource 106. Data storage device 112 stores text data, color data, style data, video data, layout data, classification data, and / or other types of data extracted from resources 106 or obtained from other data sources. there is a possibility. Data assets may be stored in the data storage device 112 in association with the resource identifier. The resource identifier may indicate the resource from which the data asset was extracted or the resource with which the data asset is associated.

  In some implementations, the data storage device 112 stores associations between landing resources and secondary resources (eg, in a database, in a data table, etc.). Content generation system 114 may access data storage device 112 to identify secondary resources associated with the specified landing resource. The data storage device 112 may store the URL of a particular social networking resource (eg, Google+ page) associated with the specified landing resource. In some implementations, the data storage device 112 stores branding information associated with the resource 106. The data storage device 112 may store a knowledge base that includes semantic information collected from a wide variety of sources (eg, Google Knowledge Graph). The knowledge base may provide structured detailed information about various landing resources, in addition to a list of links to other resources associated with the landing resource.

  The data storage device 112 may store previous content items that were used in connection with the content requester 104. Previous content items include content items provided by content requester 104, content items generated by content generation system 114 for content requester 104, images previously used or approved by content requester 104, and / or May include other components of the content item generated. The data storage device 112 is an image on a page extracted from the resource 106, an image previously used or approved by the content requester 104, and / or other that was not extracted from the resource 106 or not approved by the content requester 104 Could be an image repository for images.

  In some implementations, the data storage device 112 stores event data, statistical data, or performance data related to content items provided to the client device 108. Event data is provided with an indication of the service delivery context (e.g., device type, platform, etc.) and one or more attributes (e.g., click type, creative extension, etc.) and / or provided content item. User behavior associated with the content item (eg, click, conversion, etc.). The statistical data may include statistics (eg, total clicks, total conversions) based on event data for various content items or groups of content items. Statistical data may indicate the performance of a content item having a particular feature or set of features in a given service provisioning context. The performance data may include performance metrics (eg, CTR) calculated based on event data and / or statistical data regarding the content item or group of content items.

  With continued reference to FIG. 1, computer system 100 is shown including a content generation system 114. Content generation system 114 may be configured to extract data assets (eg, images, colors, text, fonts, styles, etc.) from resource 106. The content generation system 114 processes the extracted data assets to generate additional data assets (e.g., generate call-to-action text based on web page classification) and / or additional data assets. Can identify secondary resources from which can be obtained. The content generation system 114 processes the extracted data assets (e.g. crops images, identifies high contrast color schemes, generates text snippets from extracted text, etc.) and within the generated content item Various data assets can be selected for use.

  Content generation system 114 may generate a content item that includes the selected data asset. In some implementations, the content generation system 114 generates content items in a variety of different formats (eg, video, display, text, etc.) and stores the content items in each format. The content generation system 114 may provide the content requester 104 with the highest score content item for approval and / or store the content item for subsequent delivery to the client device 108. The content generation system 114 is described in more detail with reference to FIGS.

  Referring now to FIG. 2, a closed loop system 200 for optimizing content performance in accordance with the described implementation is shown. System 200 is shown to include content generation system 114 and client device 108. Content generation system 114 may receive content request 202 from a remote system or device. In various implementations, the content request 202 may be received from the content requester 104, the content server, the real-time bidding system, or the client device. The content request 202 may be received from the client device 108 as part of an online content provision process (eg, a request for an advertisement for display on a web page). Content request 202 may include device type, platform, location of device on which content is presented, current time, category of interest associated with client device, ID or category of resource with which content is presented, or specific impressions and It may include an indication of a specific service provision context, such as other attributes of the service provision context.

  The content generation system 114 is shown to receive event data 204 from the event log database 208. In various implementations, the event data 204 may be received from the event log database 208, received from a log file processing system, or received directly from the client device 108. Event data 204 may indicate the behavior of client device 108 in relation to a content item previously provided to client device 108. Content generation system 114 may use event data 204 to train one or more content performance prediction models 206. The content performance prediction model 206 may be configured to predict the performance of a feature or combination of features that may potentially be included in a content item generated for a given service delivery context.

  The content generation system 114 uses the predictive model 206 to select a specific feature or combination of features that is estimated to lead to optimal content performance (eg, a feature that maximizes performance metrics such as pCTR). Can do. In some implementations, the content generation system 114 uses the prediction model 206 to generate or select a creative 212, a click type 214, and / or a creative extension 216. Generating creative 212 may include extracting a data asset from the landing resource and generating a creative tailored to the landing resource using the selected data asset. Creative generation is described in more detail with reference to FIGS.

  Selecting the click type 214 uses the predictive model 206 to determine the estimated performance for content items with various click type attributes (eg, button click type, overall content item click type, confirm click type, etc.). It may include calculating a metric (eg, pCTR). The content generation system 114 may select a click type 214 from a plurality of potential click types based on the estimated performance metric. In some implementations, the content generation system 114 selects a click type 214 that optimizes the calculated performance metric for a particular service delivery context.

  Selecting creative extension 216 determines that which of a plurality of creative extensions (e.g., click-to-call button, location button, secondary resource button, etc.) is available for content request 202. May be included. The availability of the creative extension 216 can vary based on the content request 202 and the particular service delivery context. For example, the click-to-call button may be available when the client device 108 is a mobile device (eg, a smartphone), but may not be available when the client device 108 is a desktop client. Content generation system 114 may generate a creative extension using landing resources and / or other information extracted from other databases about the content item (eg, company contact information, company location, etc.).

  The content generation system 114 may generate a content item 210 having selected features for content performance optimization. The content item 210 is shown to include a selected creative 212, a selected click type 214, and a selected creative extension 216. The content generation system 114 may deliver the content item 210 to the client device 108 for presentation to the user. User actions associated with content item 210 may generate event data 204 that is stored in event log database 208 for use in further training predictive model 206.

  The closed loop system 200 allows the event data 204 to be used as feedback for training and / or updating the predictive model 206. Training the predictive model 206 using the event data 204 may increase the accuracy of the prediction, and the predictive model 206 (e.g., at regular intervals or continued when new events are logged) Potentially) being able to be updated.

  Referring now to FIG. 3, the content generation system 114 according to the described implementation is shown in more detail. Content generation system 114 is shown to include a processing circuit 302 having a processor 304 and a memory 306. The processor 304 can be a general purpose processor, application specific integrated circuit (ASIC), one or more field programmable gate arrays (FPGA), CPU, GPU, group of processing components, or other suitable electronic processing components. Can be implemented as

  Memory 306 is one or more devices (e.g., RAM) for storing data and / or computer code for completing and / or facilitating the various processes, layers, and modules described in this disclosure. , ROM, flash memory, hard disk storage, etc.). Memory 306 may include volatile memory or non-volatile memory. Memory 306 may include database components, object code components, script components, or any other type of information structure to support the various activities and information structures described in this disclosure. In some implementations, the memory 306 is communicatively connected to the processor 304 via the processing circuit 302 and is computer code (e.g., for performing one or more processes described herein). , Data modules stored in the memory 306).

  With continued reference to FIG. 3, the memory 306 is shown to include a content extractor 308, a creative selector 310, a click type selector 312, and a creative extension selector 314. Content extractor 308 and selectors 310-314 may be modules of content generation system 114 (eg, in memory 306). In various implementations, the content generation system 114 may be a single system or device (as shown in FIG. 3) or distributed across multiple systems or devices. In a distributed implementation, the functions of content extractor 308, creative selector 310, click type selector 312, and creative extension selector 314 are performed by a number of different systems or devices configured to communicate over network 102. There is a possibility that.

  In some implementations, the content extractor 308 is configured to run once per creative. Content extractor 308 identifies and extracts information that can be used to select or generate a creative (eg, by creative selector 310) or to select or generate a creative extension (eg, by creative extension selector 314) there is a possibility. Content extractor 308 may be executed prior to service delivery time to build a library of data assets and / or information that may be selected later for inclusion in content item 210. The content extracted by the content extractor 308 may determine what type of creative extension can be used for selection. The click type selector 312 and the creative extension selector 314 are at service time (i.e., in response to the content request 202) to select the best click type and / or creative extension to include in the generated content item 210. May be executed. Click type selector 312 and creative extension selector 314 may select a click type and creative extension based primarily on the context of service provision.

  Content extractor 308 may be configured to extract data assets from various data sources for use in generating content item 210. The resource from which the data asset is extracted may be defined by the identification information of the landing resource. The extracted data assets include images, colors, fonts, text, graphics, videos, styles, company related information (eg, geographic location, contact information, etc.), page classification, and / or landing resources 318. Other types of data related to

  Content extractor 308 may extract data assets from landing resources 318, secondary resources 320 associated with landing resources 318, and / or customer information database 322. In some implementations, the landing resource 318 is one of the resources 106. Landing resource 318 may be defined as a resource to which client device 108 is directed by content item 210. The secondary resource 320 can be any type of resource (eg, web page, website, document processing document, video, PDF document, etc.) and can be associated with the landing resource 318. In some implementations, secondary resource 320 is a social networking resource, such as a Google+ page associated with landing resource 318.

  Customer information database 322 may store various types of information regarding landing resources 318 and / or content requester 104. For example, customer information database 322 may provide information that can be used to generate creative extensions for inclusion in content item 210. In some implementations, the customer information database 322 includes company contact information (e.g., phone numbers, email addresses, etc.), the company's geographic location, the type of goods or services provided by the company, Store information related to the company, such as a classification or other type of information that may be included in the content item 210.

  In some implementations, content extractor 308 includes rendered resources configured to generate a rendered representation of landing resource 318 and / or secondary resource 320. The rendered representation can be a document object model (DOM) tree or a snapshot image that identifies various data assets extracted by the content extractor 308. The content extractor 308 is described in more detail with reference to FIG.

  With continued reference to FIG. 3, the data assets extracted by the content extractor 308 may be provided to the creative selector 310, click type selector 312, and creative extension selector 314. Creative selector 310 may be configured to select or generate a creative 212 for use within content item 210. In various implementations, the creative selector 310 selects a previously generated creative from the creative database 316 or uses the creative generator 324 to generate a new creative. The creative generator 324 is tailored to the landing resource 318 for presentation on the client device 108 using the data asset extracted by the content extractor 308 and / or the service delivery context provided by the content request 202. Generate creatives that are Creative generator 324 is described in more detail with reference to FIG.

  With continued reference to FIG. 3, the memory 306 is shown to include a click type selector 312. Click type selector 312 may be configured to select a click type 214 for use within content item 210. The click type 214 can be an attribute of the content item 210 or a snippet of code embedded in the content item 210. The click type 214 can affect the layout of the content item 210, the interactivity of the content item 210, and / or how the content item 210 is rendered or presented by the client device 108. In some implementations, the click type 214 defines the type of user action that qualifies as a “click” for the purpose of charging the content provider and calculating performance metrics such as CTR. The click type selector 312 may select a click type 214 from a plurality of potential click types including a “button” click type, an “entire content item” click type, and a “confirm” click type.

  A button click type may define a click as a user interaction with a particular part of the content item 210, such as a clickable button or other subset of the entire area of the content item 210. If the content item 210 has a button click type, user interaction with other parts of the content item 210 (eg, parts outside the clickable area) may not cause the client device 108 to navigate to the landing resource 318 And may not qualify as a click for purposes of calculating billing and / or performance metrics.

  The entire content item click type may define a click as a user interaction with any part of the content item 210. If the content item 210 has a whole content item click type, user interaction with any part of the content item 210 may cause the client device 108 to navigate to the landing resource 318, and billing and / or performance measurements May be eligible for a click to calculate the criteria.

  The confirmation click type may cause the content item 210 to display a confirmation prompt in response to a first user interaction with the content item 210 (eg, a first click). The confirmation prompt may display a message instructing the user to click again to navigate to the landing resource 318. If the content item 210 has a confirm click type, a second user interaction with the content item 210 (e.g., a second click) may cause the client device 108 to navigate to the landing resource 318, and charge and May be eligible as a click for the purpose of calculating performance metrics. Confirmation click types can help protect against unintentional clicks that are unlikely to lead to conversions.

  With continued reference to FIG. 3, the click type selector 312 is shown to include a click type performance model 326 and a click type model trainer 328. Click type performance model 326 may be configured to predict the performance of content item 210 for each of the potential click types that may be selected by click type selector 312 based on the service provisioning context of content item 210.

  Click type performance model 326 includes landing resource ID (e.g., URL) or type (e.g., page classification), device type (e.g., mobile, tablet, desktop, etc.), content platform (e.g., video, text, image, etc.), It may be configured to receive input such as device location, user interest, or other information indicating a particular service delivery context. The click type selector 312 may use the click type performance model 326 to calculate expected performance metrics (eg, pCTR) for various click types based on the service delivery context of the content item 210. The click type selector 312 may select the click type 214 based on expected performance metrics for various click types.

  Click type model trainer 328 may be configured to access event log database 208 to retrieve event data 204 and / or performance metrics 334. Event data 204 and performance metric 334 may indicate the performance of a content item previously provided to client device 108 (or other client device). Click type model trainer 328 uses data from event log database 208 to ensure that click type performance model 326 provides click type selector 312 with an accurate prediction of content performance for various click types. The click type performance model 326 may be trained and / or updated. Click type selector 312 is described in more detail with reference to FIG.

  With continued reference to FIG. 3, the memory 306 is shown to include a creative extension selector 314. Creative extension selector 314 may be configured to select a creative extension 216 for use within content item 210. The creative extension 216 can be a supplemental element of the content item 210 in addition to the main creative 212. Creative extension selector 314 may select a creative extension 216 from a plurality of potential creative extensions based on the service provisioning context for content item 210 and / or the data assets extracted by content extractor 308. In some implementations, potential creative extensions include clickable buttons such as “click to call” buttons, “position” buttons, “secondary resources” buttons, and / or other types that may be included in content items. Includes supplemental content.

  A click-to-call button may be used for selection when the context of service provision for the content item 210 indicates that the content item 210 is presented on a mobile communication device. In some implementations, a click-to-call button may be available for selection when a phone number or other contact information is extracted by the content extractor 308. Contact information may be automatically extracted by the content extractor 308 from the landing resource 318 or the customer information database 322. Selecting a click-to-call button may initiate a call or other communication with an entity (eg, retail store, customer service phone, sales agent, etc.) associated with the content item. For other device types (for example, desktop computers, tablets, etc.), when the contact information is known, there is a “click to email” button or a wider “click to contact” button instead of the click to call button. May be included.

  The location button is selected when the geographical location (for example, street address, store location, email address, GPS coordinates, etc.) is extracted by the content extractor 308 or when the location of the client device 108 is known May be available. Selecting the location button may cause a map showing the extracted geographic location to be displayed. If the context of service provision indicates that the content item 210 is presented on the mobile device, selecting the location button may cause the map application or navigation application to load. In some implementations, selecting a location button causes the client device 108 to navigate to a map resource (eg, a Google map).

  The secondary resource button may be used for selection when the identification information of the secondary resource 320 is known. Selecting the secondary resource button may cause the client device 108 to navigate to the secondary resource 320. Content item 210 may include creative extensions (eg, selectable buttons or links) for a number of different resources or resources (eg, secondary resources 320, social resources, landing resources 318, etc.). In some implementations, the content item 210 links only to the secondary resource 320 (not the landing resource 318) and may be referred to as a social content item. In other implementations, content item 210 links to both landing resource 318 and secondary resource 320.

  With continued reference to FIG. 3, the creative extension selector 314 is shown to include a creative extension performance model 330 and a creative extension model trainer 332. The creative extension performance model 330 is configured to predict the performance of the content item 210 for each of the available creative extensions or combinations of creative extensions that may be selected by the creative extension selector 314 based on the service provisioning context of the content item 210. Can be configured.

  Creative enhanced performance model 330 includes landing resource ID (e.g., URL) or type (e.g., page classification), device type (e.g., mobile, tablet, desktop, etc.), content platform (e.g., video, text, image, etc.), It may be configured to receive input such as device location, user interest, or other information indicating a particular service delivery context. The creative extension selector 314 uses the creative extension performance model 330 to calculate expected performance metrics (e.g., pCTR) for various types of creative extensions based on the service delivery context of the content item 210. obtain. Creative extension selector 314 may select a creative extension or a combination of creative extensions for content item 210 based on expected performance metrics.

  Creative extension model trainer 332 may be configured to access event log database 208 to retrieve event data 204 and / or performance metrics 334. Event data 204 and performance metric 334 may indicate the performance of a content item previously provided to client device 108 (or other client device). The creative extension model trainer 332 uses the data from the event log database 208 to ensure that the creative extension performance model 330 provides the creative extension selector 314 with an accurate prediction of content performance for various types of creative extensions. The creative enhanced performance model 330 may be used to train and / or update. The creative extension selector 314 is described in more detail with reference to FIG.

  Referring now to FIG. 4, the content extractor 308 according to the described implementation is shown in more detail. Content extractor 308 is shown to include a resource renderer module 402. In some implementations, rendering of resources is performed by resource renderer module 402 rather than an external resource rendering service (eg, resource renderer 110). The resource renderer module 402 may include the functionality of the resource renderer 110 described with reference to FIG. For example, the resource renderer module 402 may be able to interpret the resource 106 and generate a representation (eg, image, display, etc.) of the resource 106.

  Resource renderer module 402 may identify a particular landing resource using a URL or other indicator provided by content request 202 as part of a request to generate a content item. The resource renderer module 402 reads and interprets content (e.g., HTML, XML, image URL, etc.) and format information (e.g., CSS, XSL, etc.) marked up from the landing resource 318 and secondary resource 320 (e.g., e.g., Resources 318-320 may be rendered (according to the W3C standard). Resource renderer module 402 may generate snapshot images of resources 318-320 and / or build a DOM tree that represents resources 318-320.

  A snapshot image can be a visual representation of a resource. For example, a snapshot image for landing resource 318 can render the appearance of landing resource 318 that is presented on a user interface device (e.g., an electronic display screen, computer monitor, touch display, etc.) after rendering landing resource 318. May show. A snapshot image may include color information (eg, pixel color, brightness, saturation, etc.) and style information (eg, right angle corners, rounded edges, modern, naive, etc.) about landing resources 318 . In some implementations, the snapshot image can be an image file with any suitable file extension (eg, .jpg, .png, .bmp, etc.).

  The DOM tree can be a hierarchical model of resources. For example, the DOM tree for landing resource 318 includes image information (e.g., image URL, display position, display size, alternative text, etc.), font information (e.g., font name, size, effect, etc.), color information (e.g. For example, RGB color values, hexadecimal color codes, etc.), and text information. Resource renderer module 402 may store snapshot images and / or DOM trees for subsequent use by other modules of content generation system 114 (eg, in memory 306 or an external database). In some implementations, the resource renderer module 402 stores the snapshot image and / or DOM tree as rendered resource data 410.

  With continued reference to FIG. 4, the content extractor 308 is shown to include a secondary resource module 404. Secondary resource module 404 may be configured to identify secondary resource 320 associated with landing resource 318. In some implementations, the secondary resource module 404 accesses the resource association database 408 to identify one or more secondary resources 320 associated with the landing resource 318. Resource association database 408 stores mappings, links, associations, and / or other connections between different resources (eg, in the form of data tables, tabular databases, associative databases, etc.) Can do. In some implementations, the resource association database 408 stores branding information associated with the resources 318-320. The resource association database 408 may store semantic information and / or detailed information about various landing resources and may include a list of links to other resources associated with the landing resource. In some implementations, the resource association database 408 stores associations between the landing resources 318 and social networking resources associated with the landing resources 318 (eg, Google+ pages).

  In some implementations, the secondary resource module 404 uses the information extracted from the landing resource 318 to identify the secondary resource 320. For example, the landing resource 318 may include a link, textual description, metadata, and / or other identifying information for a particular secondary resource 320. Secondary resource module 404 determines whether secondary resource 320 is associated with the same entity as landing resource 318 (e.g., owned, operated, or maintained by the same entity as landing resource 318), or secondary resource 320 is associated with landing resource 318. Secondary resource 320 may be authenticated to determine whether it is associated with a different principal (eg, fan page, imitation, etc.).

  Secondary resource module 404 may be configured to determine the quality of secondary resource 320. If the secondary resource 320 is a social networking resource, the secondary resource module 404 may determine the quality based on the number of connections (eg, followers, links, subscribers, etc.) with other social networking resources. In some implementations, the secondary resource module 404 determines the quality of the secondary resource 320 according to the popularity (eg, number of visits, number of views, etc.) of the secondary resource 320. The secondary resource module 404 may rank or score a plurality of secondary resources associated with the landing resource 318 and provide the resource renderer module 402 with an indication of the secondary resource with the higher score or scores.

  The resource renderer module 402 may generate a rendered representation of the secondary resource 320 and store the rendered data in the rendered resource data 410. In some implementations, the resource renderer module 402 extracts secondary resource data from the secondary resource 320 and stores it. The secondary resource data may include any type of information obtained from the secondary resource 320. Secondary resource data includes company classification, company contact information (e.g., company phone number, company email, etc.), company geographical location, type of goods or services provided by the company, images, text , Colors, styles, and / or other types of information that may be extracted from secondary resources 320.

  With continued reference to FIG. 4, the content extractor 308 is shown providing rendered resource data 410 to the various extractor modules 412-424. Each of the extractor modules 412-424 may be configured to extract different types of data from the rendered resource data 410. For example, image extractor 412 may be configured to extract image data asset 426 and video extractor 414 may be configured to extract video data asset 428 and color extras. Kuta 416 may be configured to extract color data asset 430, Font Extractor 418 may be configured to extract font data asset 432, and Text Extractor 420 may be May be configured to extract text data asset 434, and may be configured to extract location extractor 422, location data asset 436 (e.g., address, GPS coordinates, etc.) Extractor 424 may be configured to extract contact data assets 438 (eg, phone numbers, email addresses, etc.). Data assets 426-438 may be stored in memory 306 or a separate database for use in generating creative 212 and / or creative extension 216.

  Referring now to FIG. 5, the creative generator 324 according to an exemplary embodiment is shown in more detail. Creative generator 324 may use data assets 426-438 extracted by content extractor 308 to generate a creative that is customized for landing resource 318. Creative selector 310 may select creative 212 from a plurality of potential creatives generated by creative generator 324. In other implementations, the creative selector 310 may select the creative 212 from a database of creatives previously generated or provided by the content provider (ie, the creative database 316).

  Creative generator 324 is shown to include an image / video module 502. The image / video module 502 may analyze the extracted image data asset 426 to detect the visual content of the image. Detecting visual content is to determine the position of prominent objects represented in the image, to determine the position of text in the image, and / or to increase the visual impact of the image on the image Determining whether it can be cropped or processed. In some implementations, the image / video module 502 analyzes the extracted image to detect the semantic content of the image. Detecting the semantic content may include identifying the meaning conveyed by the object or image shown in the image. Image / video module 502 may assign an image one or more labels or keywords that indicate the semantic content of the image. Labels and / or keywords can be used to determine the relevance of an image to a particular third party content item.

  The image / video module 502 may process the images and prepare the images for use in the third party content item. Image processing involves cropping an image to enhance noticeable objects or remove text, resize the image, format the image, or otherwise adjust the image. May be included. In some implementations, the image / video module 502 identifies and expands the logo image.

  Image / video module 502 may filter and rank images based on various attributes of the images. The image / video module 502 may determine a quality score and / or saliency score for each of the images. The quality score for an image may indicate the aesthetic appearance of the image based on various image attributes. The saliency score may indicate the saliency that the extracted image is displayed on the landing resource. The image / video module 502 may discard or filter images having a display size less than the threshold display size or a quality score less than the threshold quality score. In some implementations, the image / video module 502 ranks images based on a saliency score associated with the image. The image / video module 502 may select the image or images with the highest rank for inclusion in the display content item.

  The image / video module 502 may analyze and process the extracted video data asset 428 using processing steps similar to those applied to the image data asset 426. In some implementations, the image / video module 502 combines multiple extracted images into a video file or video slideshow.

  With continued reference to FIG. 5, the creative generator 324 is shown to include a color / font module 504. The color / font module 504 includes color data assets 430 (e.g., background colors, image colors, font colors, text colors, etc.), font data assets 432 (e.g., font faces, font sizes, etc.), and content extras. Other types of style information extracted by Kuta 308 may be received. In some implementations, the color / font module 504 selects the background, heading, description, button background, and / or button text color of the generated content item. The color / font module 504 may identify and use colors that are displayed on the landing resource 318, the secondary resource 320, or a combination thereof. The color / font module 504 may select a color for the content item using the landing resource snapshot image and / or DOM tree. In some implementations, the color / font module 504 identifies a cluster of colors from the snapshot image using a clustering technique (eg, k-means clustering).

  The color / font module 504 may select a font or font family for use in the generated content item. In some implementations, the font data asset 432 may include font information as HTML, CSS, or XML font tags. The color / font module 504 may use the font data asset 432 to identify and select one or more fonts (eg, font faces, font families, etc.).

  In some implementations, the color / font module 504 divides the extracted fonts into multiple categories based on the font size. For example, the color / font module 504 may generate a first category for large fonts (eg, greater than 20 pt, greater than 16 pt, etc.) and a second category for relatively small fonts. In some implementations, the color / font module 504 identifies multiple fonts or font families for use within the generated content item. For example, the color / font module 504 identifies a first font to use as the heading font for the generated content item and a second font to use as the font for the descriptive part of the content item or button text. there's a possibility that. The color / font module 504 is a style generator module for using the extracted colors and fonts in generating a style (eg, color scheme, font scheme, etc.) for the generated content item 210. May be offered to 510.

  With continued reference to FIG. 5, the creative generator 324 is shown to include a text module 506. The text module 506 receives the text data asset 434 and uses the text data asset 434 to generate heading text, descriptive text, metadata on the page, title on the page, caption for the image, and generated content item. Other types of text may be selected for use within. In some implementations, the text module 506 uses the text data asset 434 to generate a text summary that is displayed on the landing resource 318.

  In some implementations, the text module 506 retrieves textual data from other data sources in addition to or instead of the landing resource 318. For example, the text module 506 may receive textual data from reviews created by a company, product, or service user. Reviews may be retrieved from Internet resources (eg, websites) that allow users to post or send comments, reviews, or other text related to a particular company, product, or service. The URL of landing resource 318 may be used to specify the location of such reviews and / or direct text module 506 to a particular resource.

  The text module 506 determines whether a critique is positive or negative by a numerical rating (e.g., `` 1 out of 5 '', `` 4 stars '', etc.) or according to such a rating. It may include a sentiment detection system that can be determined anytime. The sentiment detection system may analyze critical words and look for affirmative adjectives (eg, extraordinary, good, great, nice, etc.). The emotion detection system may then select or extract a relatively short snippet of criticism that includes such a positive phrase for inclusion in the content item 210.

  With continued reference to FIG. 5, the creative generator 324 is shown to include a classification module 508. In some implementations, the classification module 508 uses the resource data 410 to classify or categorize the landing resources 318. In other implementations, the classification module 508 accesses a category database (e.g., customer information database 322 or resource association database 408) to obtain categories for the landing resource 318 (e.g., based on the landing resource 318 URL). obtain. The category database may specify a company category (eg, fast food, auto parts, etc.) associated with the landing resource 318 and other attributes of the associated company. The category of landing resource 318 may be used by text generator module 512 to generate action text for content item 210 to be generated.

  The classification module 508 may determine the classification of the landing resource 318 using data from the secondary resource 320. For example, if the secondary resource 320 is a social networking resource such as a Google+ profile, it may include classification information provided by the same subject associated with the landing resource 318. Classification module 508 may use the classification information from secondary resource 320 to determine whether landing resource 318 is generated for branding or for a company in a particular region. The classification information provided by the user may provide more detailed and accurate categories than other sources of classification information and can be advantageously used by the classification module 508 to improve the classification of landing resources 318 There is sex.

  With continued reference to FIG. 5, the creative generator 324 is shown to include a style generator module 510. The style generator module 510 may be configured to generate a color scheme and / or font scheme for the content item 210 to be generated. Style generator module 510 may use the colors extracted by color / font module 504 to generate a color scheme for automatically generated content items. The style generator module 510 may select a background color, a button color, a heading color, a description color, a button text color, or a color for other parts of the generated content item. The style generator module 510 may determine the saturation, brightness, noticeability, and / or other attributes of each extracted color and the contrast between each of the extracted colors.

  In some implementations, the style generator module 510 is the most dominant color extracted by the color / font module 504 as the background color for the content item (e.g., most heavily weighted, most dominant) Etc.). The style generator module 510 may select the extracted color with the largest number multiplied by saturation and weight as the button color for the content item. The style generator module 510 may select the color with the greatest contrast and / or brightness difference from the selected background color as the color for the heading and description text. If more than two colors are available, the style generator module 510 may select a more prominent color as the heading color.

  In other implementations, the style generator module 510 selects a pre-defined color scheme for the content item 210. Instead of immediately applying the color extracted by the color / font module 504, a pre-defined color scheme is generated, such as background color, button color, heading color, description color, button text color, or May be used to select other parts of the content item. The predefined color scheme can be a combination of colors previously grouped into a color template or group of colors. In some implementations, the predefined color scheme may be selected from a set of predefined color schemes based on the colors extracted by the color / font module 504. For example, the style generator module 510 may compare the colors extracted by the color / font module 504 with colors included in multiple predefined color schemes. The style generator module 510 determines the difference (e.g., RGB values) between one or more of the colors extracted by the color / font module 504 and one or more of the colors included in the predefined color scheme. , Saturation, brightness, contrast, etc.) may be ranked in advance. Colors from predefined color schemes may supplement or replace colors specified by the color / font module 504 in automatically generated content items.

  In some implementations, the style generator module 510 selects a color scheme that improves the readability of images and / or text. If the image is displayed within the generated content item 210, the style generator module 510 may select a background color of the image that provides high contrast with the image color (eg, RGB contrast> 200). . In some implementations, the style generator module 510 identifies the original background color of the extracted image. The original background color may be the background color of the image displayed on the resource from which the image was extracted. The style generator module 510 may use the original background color of the image in the generated content item 210.

  In some implementations, the style generator module 510 identifies a boundary in the image that separates the area within the boundary from the background color. Style generator module 510 may determine the longest boundary such that the contrast between the area within the boundary and the background color (e.g., average RGB contrast) exceeds the threshold contrast value (e.g., RGB contrast> 200) . If the area within the boundary exceeds a threshold area (eg, 80% of the image area), the style generator module 510 may determine that the color scheme improves the readability of the image.

  The style generator module 510 may select a font combination for use in the generated content item 210. In some implementations, the style generator module 510 selects a first font for headline text and a second font for descriptive text and / or button text. Style generator module 510 may use the font extracted from landing resource 318 to select the font combination that best matches the font used in landing resource 318. In this way, the style generator module 510 facilitates the generation of content items that have a look and feel similar to the landing resource 318.

  With continued reference to FIG. 5, the creative generator 324 is shown to include a text generator module 512. Text generator module 512 may be configured to generate a text portion of content item 210. Text generator module 512 may generate heading text, descriptive text, button text, action text, mouse over text, metadata text, and / or other types of text for inclusion in content item 210. .

  Text generator module 512 is shown receiving input from classification module 508 and text module 506. Text generator module 512 may generate a call to action text tailored for a particular category of landing resource 318 using the category of landing resource 318 determined by classification module 508. For example, if the landing resource 318 is associated with a game category, the text generator module 512 may generate action text such as “play now” or “challenge now”. If the landing resource 318 is associated with a subscription service, the text generator module 512 may generate a call-to-action text such as “Sign Up Now” or “Register Now”.

  In some implementations, the text generator module 512 identifies text snippets (eg, phrases, text strings, portions, etc.) for use within the generated content item 210. The text snippet may be part of text that appears on the landing resource 318, secondary resource 320, or other text that is not displayed on the resource 106 (eg, metadata text). The text generator module 512 processes various types of extracted text and can be used as descriptive text or heading text in the generated content item 210 (e.g., a threshold number of words, characters, or sentences). A short part (less than) can be identified. Text generator module 512 may highlight or emphasize important features of the text, such as price or date (eg, bold, underline, italic, increase font size, etc.).

  In some implementations, the text generator module 512 uses text extracted from user reviews of companies, products, or services associated with the landing resource 318. The text generator module 512, when read alone, selects a portion of the user's critique that effectively communicates why the user who sent the critique has had a positive experience with the critique company, product, or service Can do. The snippet may include one or more positive adjectives that are used by the text module 506 in identifying emotions related to criticism. For example, the text generator module 512 may select the snippet “extra pasta and crisp service” from a relatively long review of an Italian restaurant. In some implementations, the text snippet identified by the text generator module 512 is presented to the content requester 104 as a potential “creative” (eg, descriptive text) for use within a text-only content item. There is a possibility. In other implementations, the text snippet may be used as part of the text of one or more display content items generated by the content generation system 114.

  With continued reference to FIG. 5, the creative generator 324 is shown to include a layout module 514. The layout module 514 may be configured to generate a layout for the content item 210 using the extracted and / or processed data. In some implementations, the layout module 514 generates a creative 212 for the content item 210 and may be referred to as a creative builder. The layout module 514 may select a layout from a set of predefined layout options (eg, a template layout) or generate a new layout (eg, not based on a template). The layout module 514 may generate a layout based on the display size of the image selected by the image / video module 502 and / or the length of the text selected by the text module 506. The layout module 514 may resize the image and / or adjust the text to fit the selected layout, or adjust the layout to fit the selected image and / or text.

  In some implementations, the layout module 514 uses the data assets extracted from the landing resources 318 to determine the style, company category, or appearance for the content item 210. For example, the layout module 514 may use a landing resource 318 company category (eg, fast food, auto parts, etc.), a landing resource 318 style (eg, modern or rustic), and the shape displayed on the landing resource 318. (For example, 90 degree corners, round corners, etc.) may be determined. The layout module 514 may call an external database to retrieve company category information based on the landing resource 318 URL.

  The layout module 514 may aggregate the various data assets to generate the creative 212 and provide the creative 212 to the rendering service 336. Rendering service 336 may generate and render content item 210 in combination with creative 212 click type 214 and creative extension 216.

  Referring now to FIG. 6, the click type selector 312 according to the described implementation is shown in more detail. Click type selector 312 may be configured to select a click type 214 for use within content item 210. The click type 214 can be an attribute of the content item 210 or a snippet of code embedded in the content item 210. The click type 214 includes the layout of the content item 210 (e.g., defining what part of the content item 210 is clickable), the interactivity of the content item 210 (e.g., clicks that trigger navigation to the landing resource 318). Type) and / or how the content item 210 is rendered or presented by the client device 108. In some implementations, the click type 214 defines the type of user action that qualifies as a “click” for the purpose of charging the content provider and calculating performance metrics such as CTR. Click type 214 may define one or more conditions or triggers to cause client device 108 to navigate to landing resource 318.

  The click type selector 312 is shown to include a click type model trainer 328. Click type model trainer 328 may be configured to access event log database 208 to retrieve event data 204. Event data 204 may indicate the performance of content items previously provided to client device 108 in various service provisioning contexts. Event data 204 may include logged user actions such as clicks, conversions, and / or other user behavior associated with previously provided content items. Event data 204 includes previously provided content item attributes (e.g., click type, creative ID, creative extension, content type, etc.) and service delivery context details (e.g., device type, device location, content platform, Issuer page classification, landing resource ID, etc.). Click type model trainer 328 may train and update click type performance model 326 using data from event log database 208.

  With continued reference to FIG. 6, the click type selector 312 is shown to include a performance metric predictor 604 and a click type performance model 326. The performance metric predictor 604 may provide service provision context data 602 to the click type performance model 326. The service provision context data 602 may be received as part of the content request 202 and may indicate the context in which the content item 210 is presented. Service delivery context data 602 includes landing resource ID (e.g., URL), landing resource type (e.g., page classification), device type (e.g., mobile, tablet, desktop, etc.), content platform (e.g., video, text, image) ), The location of the device, the user profile ID associated with the content request 202, the creative ID, or other information indicating the context of the particular service offering.

  The performance metric predictor 604 uses the click type performance model 326 to calculate the predicted performance (e.g., pCTR) of content items with various click types with respect to the service delivery context indicated by the service delivery context data 602. Can do. The performance metric predictor 604 may generate a score or ranking for each click type based on the predicted performance of the click type. In some implementations, the performance metric predictor 604 predicts the performance of content items that have a “button” click type, an “all content item” click type, and a “confirm” click type.

  A button click type may define a click as a user interaction with a particular part of the content item 210, such as a clickable button or other subset of the entire area of the content item 210. If the content item 210 has a button click type, user interaction with other parts of the content item 210 (eg, parts outside the clickable area) may not cause the client device 108 to navigate to the landing resource 318. And may not qualify as a click for purposes of calculating billing and / or performance metrics.

  The entire content item click type may define a click as a user interaction with any part of the content item 210. If the content item 210 has a whole content item click type, user interaction with any part of the content item 210 may cause the client device 108 to navigate to the landing resource 318, and billing and / or performance measurements May be eligible for a click to calculate the criteria.

  The confirmation click type may cause the content item 210 to display a confirmation prompt in response to a first user interaction with the content item 210 (eg, a first click). The confirmation prompt may display a message instructing the user to click again to navigate to the landing resource 318. If the content item 210 has a confirm click type, a second user interaction with the content item 210 (e.g., a second click) may cause the client device 108 to navigate to the landing resource 318, and charge and May be eligible as a click for the purpose of calculating performance metrics. Confirmation click types can help protect against unintentional clicks that are unlikely to lead to conversions.

  With continued reference to FIG. 6, the click type selector 312 is shown to include a click type selector module 606. The click type selector module 606 may be configured to select a click type 214 from a plurality of different click types whose content performance is predicted by the performance metric predictor 604. The click type selector module 606 may consider one or more performance metrics and select a click type that optimizes such performance metrics. The click type selector module 606 may output an indication of a particular click type 214 to the rendering service 336. Rendering service 336 may combine click type 214 with creative 212 and creative extension 216 to generate and render content item 210.

  Referring now to FIG. 7, the creative extension selector 314 according to the described implementation is shown in more detail. Creative extension selector 314 may be configured to select a creative extension 216 for use within content item 210. The creative extension 216 can be a supplemental element of the content item 210 in addition to the main creative 212. Creative extension 216 may include various buttons or clickable elements that cause different actions when selected (e.g., displaying a map, initiating a call, navigating to landing resource 318, etc.) . Creative extension selector 314 may select a creative extension 314 from a plurality of potential creative extensions based on the service provisioning context for content item 210 and / or the data assets extracted by content extractor 308.

  The creative extension selector 314 is shown to include a creative extension model trainer 332. Creative extension model trainer 332 may be configured to access event log database 208 to retrieve event data 204. Event data 204 may indicate the performance of content items previously provided to client device 108 in various service provisioning contexts. Event data 204 may include logged user actions such as clicks, conversions, and / or other user behavior associated with previously provided content items. Event data 204 includes previously provided content item attributes (e.g., click type, creative ID, creative extension, content type, etc.) and service delivery context details (e.g., device type, device location, content platform, Issuer page classification, landing resource ID, etc.). Creative extension model trainer 332 may train and update creative extension performance model 330 using data from event log database 208.

  With continued reference to FIG. 7, the creative extension selector 314 is shown to include a performance metric predictor 704 and a creative extension performance model 330. The performance metric predictor 704 may provide service provision context data 602 to the creative enhanced performance model 330. The service provision context data 602 may be received as part of the content request 202 and may indicate the context in which the content item 210 is presented. Service delivery context data 602 includes landing resource ID (e.g., URL), landing resource type (e.g., page classification), device type (e.g., mobile, tablet, desktop, etc.), content platform (e.g., video, text, image) ), The location of the device, the user profile ID associated with the content request 202, the creative ID, or other information indicating the context of the particular service offering.

  The performance metric predictor 704 uses the creative extension performance model 330 to predict the expected performance (e.g., pCTR) of content items having various types of creative extensions with respect to the service delivery context indicated by the service delivery context data 602. Can be calculated. The performance metric predictor 704 may generate scores or rankings for various creative extensions based on the predicted performance of the creative extension. In some implementations, the performance metric predictor 704 has a “click to call” button, a “position” button, a “secondary resource” button, and / or other types of creative extensions that may be included in the content item 210. Predict the performance of content items.

  A click-to-call button may initiate a call or other communication with an entity (eg, a retail store, customer service phone, sales agent, etc.) associated with the content item. For other device types (for example, desktop computers, tablets, etc.), when the contact information is known, there is a “click to email” button or a wider “click to contact” button instead of the click to call button. May be included.

  The location button may display a map showing the extracted geographic location. If the context of service provision indicates that the content item 210 is presented on the mobile device, selecting the location button may cause the map application or navigation application to load. In some implementations, selecting a location button causes the client device 108 to navigate to a map resource (eg, a Google map).

  The secondary resource button may cause the client device 108 to navigate to the secondary resource 320. Content item 210 may include creative extensions (eg, selectable buttons or links) for a number of different resources or resources (eg, secondary resources 320, social resources, landing resources 318, etc.). In some implementations, the content item 210 links only to the secondary resource 320 (not the landing resource 318) and may be referred to as a social content item. In other implementations, content item 210 links to both landing resource 318 and secondary resource 320. The performance metric predictor 704 may calculate the predicted performance for content items with various types of creative extensions based on the context in which the content item is presented.

  With continued reference to FIG. 7, the creative extension selector 314 is shown to include an extension selector module 706. In some implementations, the enhanced selector module 706 receives position data 436. Location data 436 may be extracted from various data sources, such as landing resource 318, secondary resource 320, customer information database 322, or received from client device 108. Location data 436 may include a physical or geographical location of a company or entity associated with content item 210. For example, if content item 210 is associated with a restaurant chain, location data 436 may include the location of one or more of the restaurants. In some implementations, the location data 436 includes the location of the client device 108 (eg, GPS coordinates, WiFi based location, etc.).

  In some implementations, the enhanced selector module 706 receives contact data 438. Contact data 438 may be extracted from various data sources such as landing resources 318, secondary resources 320, customer information database 322, and the like. Contact data 438 may include contact information regarding a company or other entity associated with content item 210. Contact data 438 may be used to contact or communicate with a telephone number, email address, web page URL (eg, URL for landing resource 318 or secondary resource 320), or company or entity associated with content item 210 Other types of information that may be used may be included.

  The extension selector module 706 may determine which of a plurality of creative extensions may be utilized for selection using the service provision context data 602, the location data 436, and / or the contact data 438. Different types of creative extensions may be available under different circumstances. For example, a click-to-call button may be used for selection when the context of service provision for the content item 210 indicates that the content item 210 is presented on a mobile communication device. In some implementations, a click-to-call button may be available for selection when the contact data 438 includes a phone number or other contact information. The location button is selected when the geographical location (for example, street address, store location, email address, GPS coordinates, etc.) is extracted by the content extractor 308 or when the location of the client device 108 is known May be available. The secondary resource button may be used for selection when the identification information of the secondary resource 320 is known.

  The extension selector module 706 may be configured to select a creative extension or combination of creative extensions from a set of creative extensions whose content performance is predicted by the performance metric predictor 704. In some implementations, the performance metric predictor 704 predicts the performance of the content item with respect to various types of creative extensions that may be utilized for selection. The extension selector module 706 may consider one or more performance metrics and select one or more creative extensions that optimize such performance metrics. Creative extension selector module 706 may output an indication of a particular creative extension 216 or combination of creative extensions 216 to rendering service 336.

  With continued reference to FIG. 7, the rendering service 336 may combine the creative extension 216 with the creative 212 and the click type 214 to generate and render the content item 210. In some implementations, rendering service 336 is a component of client device 108 (eg, a web browser running on client device 108). In other implementations, the rendering service 336 may be a component of the content generation system 114. Each of creative 212, click type 214, and creative extension 216 may include rendering instructions specific to the selected creative, click type, or creative extension. For example, the click type 214 may render a subset of the content item 210 as a clickable element with respect to the button click type, or an instruction to render the entire area of the content item 210 as a clickable element with respect to the entire content item click type. May be included. Rendering service 336 may be configured to render content item 210 according to instructions provided by creative 212, click type 214, and creative extension 216.

  Content item 210 may be provided to client device 108. User interactions with content item 210 (eg, impressions, clicks, conversions, etc.) may be logged and stored in event log database 208. Content generation system 114 may use the logged events to calculate performance metrics associated with content item 210 and train predictive models 326 and 330. The content generation system 114 uses the predictive models 326 and 330 to determine which feature or combination of features (e.g., click type, creative extension, etc.) of the content item for optimal content performance with respect to a given service delivery context. It can be determined that it is estimated to be connected (eg, a feature that optimizes CTR, pCTR, or another performance metric). The content generation system 114 may then automatically generate new content items with features that are estimated to lead to optimal content performance.

  Referring now to FIG. 8, a flow diagram of a process 800 for optimizing content performance using click types is shown in accordance with the described implementation. Process 800 may be performed by content generation system 114 described with reference to FIGS. In some implementations, process 800 is performed by processing circuit 302 using one or more of the modules stored in processor 304 and memory 306. In various implementations, the steps of process 800 may be rearranged (ie, performed in a different order) or performed simultaneously.

  Process 800 is shown to include receiving a request for a content item for presentation on a client device (step 802). The request for the content item may include an indication of the context of service provision for the content item. Service delivery contextual instructions include landing resource ID (e.g., URL), landing resource type (e.g., page classification), device type (e.g., mobile, tablet, desktop, etc.), content platform (e.g., video, text, Service provision context data such as device location, user interest, or other information indicating a particular service provision context. A request for a content item may be received from a content requester 104, a content server, a real-time bidding system, or a client device. In some implementations, a request for a content item is received from the client device 108 as part of an online content provision process (eg, a request for an advertisement for display on a web page).

  With continued reference to FIG. 8, the process 800 uses the click type performance model and the service delivery context for the content item to calculate the expected performance metrics for the content item for a number of different potential click types. (Step 804). In some implementations, the performance metric is click rate. Each potential click type may define different criteria for defining clicks for the purpose of calculating click rates. For example, a potential click type is a button that defines a click as a user interaction with a first part of a content item excluding the user's interaction with a second part of the content item that is separate from the first part. May include a click type. A potential click type may include an entire content item click type that defines a click as a user interaction with any portion of the content item. Potential click types may include a confirmation click type that defines a click as a user interaction with a content item that occurs while a confirmation prompt is displayed. A confirmation prompt may be displayed in response to previous user interaction with the content item.

  Each potential click type may define a different criteria for triggering an event that occurs in response to a user interaction with the content item when the defined criteria are met. For example, each of the potential click types may be different user actions related to the content item (e.g., clicking a button, clicking somewhere in the content item display area, responding to a confirmation prompt) Click on the item). Triggered events can include landing resources, navigating to secondary resources, initiating communication, displaying a map, launching an application, or any other event that can be automatically triggered There is sex. Different click types may require different types of user interaction to trigger the event.

  With continued reference to FIG. 8, the process 800 is shown to include selecting one of the potential click types based on an expected performance metric (step 806). Step 806 may include selecting a click type associated with the extreme value of the predicted performance metric. In some implementations, step 806 includes selecting a button click type. If the button click type is selected, the process 800 may include user interaction with the first part of the content item (excluding user interaction with a second part of the content item that is separate from the first part). Defining criteria for triggering the event as and rendering a clickable button that occupies only the first portion of the content item.

  In some implementations, step 806 includes selecting the entire content item click type. If the entire content item click type is selected, the process 800 can define criteria for triggering an event as a user interaction with any part of the content item and can click the entire display area of the content item Rendering as an element.

  In some implementations, step 806 includes selecting a confirmation click type. If the confirmation click type is selected, the process 800 displays a confirmation prompt in response to the first user interaction with the content item, and a confirmation prompt is displayed after the first user interaction. Defining a criterion for triggering the event as a second user interaction with the content item occurring during.

  With continued reference to FIG. 8, the process 800 generates a content item having the selected click type (step 808) and provides the generated content item for presentation on the client device (step 810). ). Content items may be generated in response to the request received at step 802 as part of the real-time service provisioning process. Step 808 may include combining the selected click type with the selected creative and / or the selected creative extension. In some implementations, step 808 includes rendering the content item according to the instructions provided by the selected click type. Rendering the content item may be performed by a remote rendering service or by a client device. In some implementations, step 810 is performed before step 808. For example, the selected click type may be provided to the client device, and the client device may render the content item using the selected click type.

  With continued reference to FIG. 8, the process 800 is shown to include collecting and storing event data indicative of user actions associated with content items having each of the potential click types (step 812). The event data may indicate the performance of content items previously provided to the client device or other client devices in the context of various service provisions. The event data may include logged user actions such as clicks, conversions, and / or other user behavior associated with previously provided content items. Event data includes previously provided content item attributes (e.g., click type, creative ID, creative extension, content type, etc.) and service delivery context details (e.g., device type, device location, content platform, publishing) Pager classification, landing resource ID, etc.).

  With continued reference to FIG. 8, process 800 is shown including using event data as feedback to train a click-type performance model (step 814). Step 814 may be performed by the click type model trainer 328 described with reference to FIGS. 3 and 6. In some implementations, step 814 includes accessing an event log database (eg, database 208) to retrieve event data and / or performance metrics. Step 814 uses data from the event log database to train and / or update the click type performance model to ensure that the click type performance model accurately predicts the performance of the content for various click types. Can include.

  Referring now to FIG. 9, a flow diagram of a process 900 for optimizing content performance using creative extensions in accordance with the described implementation is shown. Process 900 may be performed by content generation system 114 described with reference to FIGS. In some implementations, process 900 is performed by processing circuit 302 using one or more of the modules stored in processor 304 and memory 306. In various implementations, the steps of process 900 may be rearranged (ie, performed in a different order) or performed simultaneously.

  Process 900 is shown to include receiving a request for a content item for presentation on a client device (step 902). The request for the content item may include an indication of the context of service provision for the content item. Service delivery contextual instructions include landing resource ID (e.g., URL), landing resource type (e.g., page classification), device type (e.g., mobile, tablet, desktop, etc.), content platform (e.g., video, text, Service provision context data such as device location, user interest, or other information indicating a particular service provision context. A request for a content item may be received from a content requester 104, a content server, a real-time bidding system, or a client device. In some implementations, a request for a content item is received from the client device 108 as part of an online content provision process (eg, a request for an advertisement for display on a web page).

  With continued reference to FIG. 9, the process 900 uses the creative extension performance model and service delivery context for the content item to calculate the expected performance metric of the content item for multiple different potential creative extensions. (Step 904). In some implementations, the performance metric is click rate. Each potential creative extension may define a different action to be taken in response to the user's interaction with the content item. For example, a potential creative extension may include a click-to-call extension that causes a client device to initiate communication with a subject associated with the content item. Potential creative extensions may include location extensions that launch a map application or display a geographic location associated with a content item. Potential creative extensions may include secondary resource extensions that cause the client device to navigate to secondary resources associated with the content item.

  With continued reference to FIG. 9, process 900 is shown to include selecting one of the potential creative extensions based on an expected performance metric (step 906). Step 906 may include selecting a creative extension associated with the extreme value of the predicted performance metric. In some implementations, step 906 includes determining which of the potential creative extensions can be utilized for selection using a service-providing context for the content item. For example, the click-to-call extension may be used for selection if the client device is identified as a mobile phone, but may not be used for selection if the client device is specified as a desktop computer.

  In some implementations, step 906 includes using the identification information of the creative portion of the content item to determine which of the potential creative extensions can be utilized for selection. For example, a “location” creative extension may be used for selection with respect to content items specifically selling a restaurant chain and may provide the location of the nearest restaurant location in the chain. However, location creative extensions may only be sold online and may not be available for selection, especially for content items that sell products or services that do not have a physical location that users may potentially visit. is there.

  In some implementations, step 906 includes extracting data assets from landing resources, secondary resources associated with the landing resources, and / or customer information database. Step 906 may include using the extracted data asset to determine which of the potential creative extensions can be utilized for selection. Determining which of the potential creative extensions can be used for selection is to identify the data assets required for the creative extension and whether the identified data assets have been extracted. Determining and in response to determining that the required data asset has not been extracted, determining that the creative extension cannot be utilized for selection.

  With continued reference to FIG. 9, the process 900 generates a content item having the selected creative extension (step 908) and provides the generated content item for presentation on the client device (step 910). ). Content items may be generated in response to the request received in step 902 as part of the real-time service provisioning process. Step 908 may include combining the selected creative extension with the selected creative and / or the selected click type. In some implementations, step 908 includes rendering the content item according to instructions provided by the selected creative extension (eg, rendering the creative extension as a selectable button). Rendering the content item may be performed by a remote rendering service or by a client device. In some implementations, step 910 is performed before step 908. For example, the selected creative extension may be provided to the client device, and the client device may render the content item using the selected creative extension.

  With continued reference to FIG. 9, process 900 is shown to include collecting and storing event data indicative of user actions associated with content items having each of the potential creative extensions (step 912). The event data may indicate the performance of content items previously provided to the client device or other client devices in the context of various service provisions. The event data may include logged user actions such as clicks, conversions, and / or other user behavior associated with previously provided content items. Event data includes previously provided content item attributes (e.g., click type, creative ID, creative extension, content type, etc.) and service delivery context details (e.g., device type, device location, content platform, publishing) Pager classification, landing resource ID, etc.).

  With continued reference to FIG. 9, process 900 is shown to include using event data as feedback to train a creative enhanced performance model (step 914). Step 914 may be performed by the creative extension model trainer 332 described with reference to FIGS. 3 and 7. In some implementations, step 914 includes accessing an event log database (eg, database 208) to retrieve event data and / or performance metrics. Step 914 trains and / or trains the creative extension performance model using data from the event log database to ensure that the creative extension performance model accurately predicts the performance of the content for various types of creative extensions. May include updating.

  Implementations and operations of the subject matter described herein include digital electronic circuits, or computer software, firmware, or hardware, or any of them, including the structures disclosed herein and their structural equivalents. It can be implemented in one or more combinations. Implementations of the subject matter described herein include one or more computer programs, ie, one or more computer storage media, for execution by a data processing device or for controlling the operation of a data processing device It may be implemented as one or more modules of computer program instructions encoded above. Alternatively or additionally, the program instructions may be artificially generated propagation signals (e.g., generated to encode information for transmission to a suitable receiver device for execution by the data processing device (e.g. , Machine-generated electrical signals, optical signals, or electromagnetic signals). The computer storage medium may be or be included in a computer readable storage device, a computer readable storage substrate, a random or serial access memory array or device, or one or more combinations thereof. Further, a computer storage medium may be a source or destination of computer program instructions that are not propagation signals but are encoded in artificially generated propagation signals. A computer storage medium may be or be included in one or more separate components or media (eg, multiple CDs, disks, or other storage devices). Thus, computer storage media is tangible and non-transitory.

  The operations described in this disclosure may be implemented as operations performed by a data processing device on data stored in one or more computer-readable storage devices or received from other sources.

  The terms “client” or “server” refer to all types of data processing, including, by way of example, one programmable processor, one computer, one system on chip, or multiple or combinations thereof. Includes devices, devices, and machines. The device may include dedicated logic circuitry, such as a field programmable gate array (FPGA) or application specific integrated circuit (ASIC). In addition to hardware, the device generates code that generates an execution environment for the computer program in question (e.g., processor firmware, protocol stack, database management system, operating system, cross-platform runtime environment, virtual machine, or A code constituting a combination of one or more of them). The device and execution environment may implement a variety of different computing model infrastructures, such as a web services infrastructure, a distributed computing infrastructure, and a grid computing infrastructure.

  The systems and methods of the present disclosure can be completed by any computer program. A computer program (also known as a program, software, software application, script, or code) can be written in any form of programming language, including a compiled or interpreted language, a declarative or procedural language, It can be arranged in any form, including as a stand-alone program or as a module, component, subroutine, object, or other unit suitable for use in a computing environment. A computer program may correspond to a file in a file system, but this is not always the case. A program is part of a file that holds other programs or data (for example, one or more scripts stored in a markup language document), a single file dedicated to the program in question, or multiple In an organized file (eg, a file that stores one or more modules, subprograms, or portions of code). The computer program may be located on one computer or on a plurality of computers that may be located at one location or distributed across multiple locations and interconnected by a communication network.

  The process and logic flow described herein includes one or more programmable programs for one or more programmable programs to perform operations by performing operations on input data and generating output. It can be executed by executing. Processes and logic flows may be performed by dedicated logic circuits (eg, FPGA or ASIC), and devices may be implemented as dedicated logic circuits (eg, FPGA or ASIC).

  Suitable processors for the execution of computer programs include, by way of example, both general and special purpose microprocessors and any one or more processors of any kind of digital computer. Generally, a processor will receive instructions and data from a read-only memory or a random access memory or both. The essential elements of a computer are a processor for performing actions in accordance with instructions and one or more memory devices for storing instructions and data. Also, generally a computer includes or receives data from one or more mass storage devices (e.g., magnetic disk, magneto-optical disk, or optical disk) for storing data. Or operably coupled to transfer data to those mass storage devices or both. However, the computer may not include such a device. In addition, the computer may be connected to another device (e.g., mobile phone, personal digital assistant (PDA), mobile audio or video player, game console, global positioning system (GPS) receiver, or portable storage device (e.g., universal serial bus). (USB flash drive) etc.). Suitable devices for storing computer program instructions and data include, by way of example, semiconductor memory devices (e.g., EPROM, EEPROM, and flash memory devices, magnetic disks, e.g., internal hard disks or removable disks, magneto-optical disks, and CDs). -All forms of non-volatile memory, media, and memory devices, including -ROM disks and DVD-ROM disks). The processor and memory may be supplemented by, or incorporated in, dedicated logic circuitry.

  In order to provide user interaction, the implementations of interest described herein include display devices (e.g., CRT (CRT), LCD (Liquid Crystal Display), OLED (Organic Light Emitting Diode), TFT (Thin Film Transistor), Or any other flexible configuration, or any other monitor that displays information to the user) and a keyboard, pointing device such as a mouse, trackball, etc., or touch screen, touch that allows the user to provide input to the computer It may be mounted on a computer having a pad or the like. Other types of devices may be further used to interact with the user, for example, the feedback provided to the user may be any form of sensory feedback (e.g. visual feedback, audio feedback) Or tactile feedback), and input from the user may be received in any form, including acoustic, speech, or tactile input. In addition, the computer sends a document to the device used by the user and receives the document from the device, for example, in response to a request received from the web browser, the web page on the web browser of the user's client device May interact with the user by sending

  Implementations of the subject matter described in this disclosure may include a back-end component (eg, as a data server) or a middleware component (eg, an application server), or a user may Includes a front-end component (e.g., a client computer) with a graphical user interface or web browser that can interact with the implementation, or one or more such back-end components, middleware components, or front-end components May be implemented on a computing system that includes any combination of: The components of the system can be interconnected by any form or medium of digital data communication (eg, a communication network). Examples of communication networks include LANs and WANs, internetworks (eg, the Internet), and peer-to-peer networks (eg, ad hoc peer-to-peer networks).

  This specification includes many specific implementation details, but these should not be construed as limitations on the scope of any disclosure or what may be claimed, but rather specific implementations of a particular disclosure. Should be regarded as an explanation of the features unique to Certain features that are described in this disclosure in connection with different implementations may also be implemented in combination in a single implementation. Conversely, various features that are described in connection with a single implementation may be implemented in multiple implementations separately or in any suitable subcombination. In addition, a feature may be described above as operating in a particular combination and may even be so claimed initially, but one or more of the claimed combinations Features may be deleted from the combination in some cases, and the claimed combination may be targeted to a partial combination or a variation of a partial combination.

  Similarly, operations are shown in a particular order in the figures, but this may be performed in the particular order shown or in sequential order, or to achieve the desired result. Therefore, it should not be understood that all operations shown for this need to be performed. In certain situations, multitasking and parallel processing may be advantageous. Further, the division of the various system components in the implementations described above should not be understood as requiring such divisions in all implementations, and the described program components and systems are generally a single unit. It should be understood that a software product may be integrated together or packaged into multiple software products embodied on one or more tangible media.

Features disclosed herein include processing circuitry configured to integrate Internet connectivity with older television program sources (e.g., received by cable, satellite, propagation, or other signals). It may be implemented on smart TV modules (or connected television modules, hybrid TV modules, etc.) that may be included. Smart TV modules may be physically integrated into a TV set or such as set-top boxes, Blu-ray or other digital media players, game consoles, hotel TV systems, and other companion devices May include separate devices. Smart TV module to allow viewers to search and find videos, movies, photos, and other content stored on the web, local cable TV channels, satellite TV channels, or local hard drives May be configured. Set-top boxes (STBs) or set-top units (STUs) contain tuners and may contain information appliance devices that connect to TV sets and external sources of signals, converting signals to content. And the content is displayed on a television screen or other display device. Smart TV modules are icons for multiple different applications such as web browsers, and multiple streaming media services (e.g. Netflix, Vudu, Hulu, etc.), connected cable or satellite media sources, other web "channels" Can be configured to provide a home screen or top-level screen that includes The smart TV module may be further configured to provide an electronic program guide to the user. A companion application for a smart TV module provides mobile users with additional information about available programs, allows users to control the smart TV module, etc. It may be possible to operate on the device. In alternative embodiments, the features may be implemented on a laptop or personal computer, a smartphone, other mobile phones, handheld computers, tablet PCs, or other computing devices.

  Thus, a specific implementation of interest has been described. Other implementations are within the scope of the appended claims. In some cases, the actions recited in the claims can be performed in a different order and still achieve desirable results. In addition, the processes shown in the accompanying drawings do not necessarily have to be in the specific order or sequential order shown in order to achieve the desired result. In certain implementations, multitasking and parallel processing may be advantageous.

100 computer system
102 network
104 Content Requester
106 Resources
108 Client devices
110 Resource Renderer
112 Data storage devices
114 Content generation system
200 systems
202 Content request
204 Event data
206 Content Performance Prediction Model
208 Event log database
210 Content items
212 creative
214 Click type
216 Creative extension
302 Processing circuit
304 processor
306 memory
308 Content Extractor
310 Creative selector
312 Click type selector
314 Creative extension selector
316 Creative Database
318 Landing resources
320 Secondary resource
322 Customer information database
324 Creative Generator
326 Click type performance model
328 Click type model trainer
330 Creative Extended Performance Model
332 Creative Extended Model Trainer
334 performance metrics
336 rendering services
402 Resource Renderer Module
404 Secondary resource module
408 Resource Association Database
410 Rendered resource data
412 Image Extractor
414 Video Extractor
416 color extractor
418 Font Extractor
420 Text Extractor
422 position extractor
424 Contact Extractor
426 Image Data Asset
428 Video Data Asset
430 color data assets
432 Font data assets
434 Text Data Asset
436 Location data assets
438 Contact Data Asset
502 Image / Video module
504 color / font module
506 text module
508 classification module
510 style generator module
512 text generator module
514 Layout module
602 Service provision context data
604 Performance Metric Predictor
606 Click type selector module
704 Performance Metric Predictor
706 Extended selector module
800 processes
900 process

Claims (20)

A method for optimizing content performance,
Receiving at a processing circuit a request for a content item for presentation on a client device, the request including an indication of a context of service provision for the content item;
Using the click type performance model for the content item and the service provision context to calculate a predicted performance metric of the content item for a plurality of different potential click types by the processing circuit. Wherein each of the potential click types defines a different criterion for triggering an event that occurs in response to a user interaction with the content item when a defined criterion is met. When,
Selecting one of the potential click types by the processing circuit based on the predicted performance metric;
Generating a content item having the selected click type, wherein the content item is generated by the processing circuit in response to the request;
Providing the generated content item for presentation on the client device. Collecting and storing event data indicative of user actions associated with content items having each of the potential click types;
2. The method of claim 1, further comprising using the event data as feedback to train the click type performance model.   The method of claim 1, wherein the different criteria defined by the potential click types include different types of user interactions with the content item.   The method of claim 1, wherein each of the potential click types defines a different criterion for calculating the performance metric. Selecting one of the potential click types comprises selecting a button click type, the method comprising:
Defining the criteria for triggering the event as a user interaction with the first part of the content item, excluding a user interaction with a second part of the content item separate from the first part And steps to
2. The method of claim 1, further comprising rendering a clickable button that occupies only the first portion of the content item. Selecting one of the potential click types comprises selecting an entire content item click type, the method comprising:
Defining the criteria for triggering the event as a user interaction with any portion of the content item;
2. The method of claim 1, further comprising: rendering the entire display area of the content item as a clickable element. Selecting one of the potential click types comprises selecting a confirmation click type, the method comprising:
Displaying a confirmation prompt in response to a first user interaction with the content item;
Defining the criteria for triggering the event as a second user interaction with the content item that is performed after the first user interaction while the confirmation prompt is displayed; and The method of claim 1 comprising:   The method of claim 1, wherein the service provisioning context includes at least one of the client device type, the client device platform, and the resource with which the content item is presented in association.   The method of claim 1, wherein selecting one of the potential click types includes selecting a click type associated with an extreme value of the predicted performance metric for the potential click type. The method described. Comparing the predicted performance metric with a threshold;
Selecting one of the potential click types comprises:
Selecting a first click type of the potential click types if the predicted performance metric is less than the threshold;
2. The method of claim 1, comprising: selecting a second click type of the potential click types when the predicted performance metric is greater than or equal to the threshold. A system for optimizing content performance,
A processing circuit having a processor and a memory, the processing circuit comprising:
Receiving a request for a content item for presentation on a client device, wherein the request includes an indication of a context of service provision for the content item;
Using the click type performance model for the content item and the service provisioning context to calculate a measure of the expected performance of the content item for a plurality of different potential click types, Defining and using different criteria for triggering an event that occurs in response to a user interaction with the content item when each defined click type is met,
Selecting one of the potential click types based on the predicted performance metric;
Generating a content item having the selected click type in response to the request;
A system configured to provide the generated content item for presentation on the client device. The processing circuit is
Collecting and storing event data indicating user actions associated with content items having each of the potential click types;
The system of claim 11, configured to use the event data as feedback for training the click-type performance model.   12. The system of claim 11, wherein the different criteria defined by the potential click types include different types of user interactions with the content item.   The system of claim 11, wherein each of the potential click types defines a different criterion for calculating the performance metric. Selecting one of the potential click types includes selecting a button click type, the processing circuit comprising:
Defining the criteria for triggering the event as a user interaction with the first part of the content item, excluding a user interaction with a second part of the content item separate from the first part And
12. The system of claim 11, configured to render a clickable button that occupies only the first portion of the content item. Selecting one of the potential click types includes selecting an entire content item click type, the processing circuit comprising:
Defining the criteria for triggering the event as a user interaction with any part of the content item;
The system of claim 11, configured to render the entire display area of the content item as a clickable element. Selecting one of the potential click types includes selecting a confirmation click type, the processing circuit comprising:
Configuring the content item to display a confirmation prompt in response to a first user interaction with the content item;
Configured to define the criteria for triggering the event as a second user interaction with the content item that occurs while the confirmation prompt is displayed after the first user interaction. The system according to claim 11.   12. The system of claim 11, wherein the service provisioning context includes at least one of the client device type, the client device platform, and the resource with which the content item is presented in association.   12. The method of claim 11, wherein selecting one of the potential click types includes selecting a click type associated with an extreme value of the predicted performance metric for the potential click type. The system described. The processing circuit is
Configured to compare the predicted performance metric to a threshold;
Selecting one of the potential click types;
Selecting a first click type of the potential click types if the predicted performance metric is less than the threshold;
12. The system of claim 11, comprising selecting a second click type of the potential click types when the predicted performance metric is greater than or equal to the threshold.

Images