In general, in one aspect, a method includes compiling user interaction statistics for a set of content items displayed in association with a first target media document having a non-textual portion, at least some of the content items associated with one or more keywords, based on the interaction statistics, associating the first target media document with at least some of the keywords associated with the content items, and based on a common attribute of the first target media document and a second target media document having a non-textual portion, associating the second target media document with at least some of the keywords assigned to the first target media document. Other aspects include corresponding systems, apparatus, and computer programs stored on computer storage devices.

A computer-implemented method can facilitate delivery of targeted content to user devices in situations in which historic tracking data (e.g., cookie data) is generally unavailable and/or unreliable. A p-dimensional embedding of websites can be generated based on a group of user devices for whom tracking data is available. Conversion event data that indicates indicating whether that audience member performed a conversion action can be received. A machine learning model can be trained using the conversion event data and the positions of websites appearing in the conversion event data within the p-dimensional embedding to predict a likelihood of conversion and/or a type of content to provide given a position in the p-dimensional embedding. When an indication that a user device is accessing a website is received, a position of that website in the p-dimensional embedding can be determined and targeted content can be delivered to the user device.

A computer-implemented method can facilitate delivery of targeted content to user devices in situations in which historic tracking data (e.g., cookie data) is generally unavailable and/or unreliable. A p-dimensional embedding of websites can be generated based on a group of user devices for whom tracking data is available. Conversion event data that indicates indicating whether that audience member performed a conversion action can be received. A machine learning model can be trained using the conversion event data and the positions of websites appearing in the conversion event data within the p-dimensional embedding to predict a likelihood of conversion and/or a type of content to provide given a position in the p-dimensional embedding. When an indication that a user device is accessing a website is received, a position of that website in the p-dimensional embedding can be determined and targeted content can be delivered to the user device.

Methods and apparatuses that utilize machine learning techniques to dynamically adjust the placement of secondary content that is displayed across numerous user devices over time are described. The user devices may comprise electronic computing devices, such as a mobile phones and digital televisions. The secondary content may be displayed within open slots of webpages or display screens in response to being selected for display during a real-time bidding process for the open slots. In some cases, in response to a bid request for an open slot within a webpage or display screen, a computer-implemented bid generation system for determining the selection and placement of secondary content may identify the secondary content to be displayed within the open slot, determine a bid amount for the identified secondary content, and transmit a bid response that includes the bid amount and the identified secondary content.

A promotion information pushing method is disclosed, including: obtaining promotion information that needs to be pushed and user behavior data; determining a degree of relevance between a user and the promotion information, and a personal impact degree and a social impact degree of the user according to the user behavior data; determining a pushing target of the promotion information based on the degree of relevance between the user and the promotion information, the personal impact degree, and the social impact degree; and pushing the promotion information to the pushing target. In addition, a corresponding apparatus and a corresponding system are provided.

A recommender engine is configured to access memory and surface transmedia content items; and/or linked transmedia content subsets; and/or one or more identifications of identified users; and/or content items of the plurality of transmedia content items associated with at least one identified user. The surfaced items are presented for selection by the given user via the transmedia content linking engine as one or more user-selected transmedia content items.

The current document is directed to methods and systems that receive instrumentation-generated events and that employ statistical inference to discover event topics and to assign an action to each of a number of events and that use the actions to predict future events and actions. In a described implementation, accumulated action messages are used to build a predictive model for each monitored website and the predictive model is used, in turn, to predict future actions based on already received actions.

A digital brand asset system is provided enabling a brand owner to create, distribute, maintain, manage, merchandise and analyze smart brand assets. The system enables distribution and sharing of smart brand assets across the websites. The websites can host webpages containing codes representing the smart brand assets. When a user device retrieves a webpage from one of the websites and renders the webpage, it executes the codes and requests the content of the smart brand assets from a brand asset server. Through the brand asset server, a brand owner can control the content and the presentation of the smart brand asset hosted by the websites, based on various factors such as previous click through rates, aggregated shopper behaviors, geographical locations of the websites or website visitors, categorized types of websites, blacklist of websites.

Acquiring first customer image information related to a photographed image of a customer when entering a store, acquiring second customer image information related to a photographed image of the customer when advancing into an accounting area, acquiring third customer image information related to a photographed image of the customer leaving the accounting area, acquiring purchase information related to a product purchase status of a customer, measuring an excursion time for each customer on the basis of the first and second customer image information, correlating the excursion time and purchase information for each customer on the basis of the third customer image information, and generating analysis information in which the excursion time and purchase information for each customer are collected, and presenting the analysis information to a user.

Systems and methods for determining advertisement content based on cluster density analysis of advertisement targets within a dynamic visibility fence are disclosed. A computer-implemented method includes determining, by an advertisement determination device, a dynamic visibility fence for a time, the dynamic visibility fence defining a first set of advertisement targets, determining, by the advertisement determination device, an advertisement to display based on cluster analysis of the first set of advertisement targets, and displaying, by the advertisement determination device, the determined advertisement.