Abstract: A digital magazine server generates a digital magazine for user based on a received request for the digital magazine identifying one or more topics. The digital magazine server applies one or more machined trained models to obtained content items to select content items for the topic. A hierarchy of the topics included in the received request may be determined by the digital magazine server and used by the trained models to select content items. When generating the digital magazine, the digital magazine server also includes one or more editorial content items that are manually selected. The digital magazine serer may reposition one or more content items selected by the trained models to include an editorial content items.

Abstract: A digital magazine server generates a model to associate topics with content items. To generate the model, the digital magazine server selects a set of content items that have been included in one or more digital magazines. For each content item of the set, the digital magazine server determines a distribution of concepts associated with a content item of the set based on characteristics of digital magazines including the content item of the set and a distribution of topics associated with the content item of the set based on words included in the content item. Additionally, the digital magazine server determines a parameter defining a relationship between the distributions of concepts and the distributions of topics associated with content items of the set. A model based on the distributions of concepts and of topics as well as the parameter is generated and stored for application to content items.

Abstract: A digital magazine server displays content items from various sources to users of client devices. Each source of a content item is identified by a domain, and content items for different sources have different domain-level quality. To differentiate sources of content items, the domains identifying the sources are ranked based on domain scores of the domains generated by an aggregate of multiple trained domain classifiers. A domain score of a domain indicates a domain-level quality of content items provided by a source identified by the domain. Each of the trained domain classifiers (e.g., a naïve Bayes classifier, a random forest classifier, and a logistic regression classifier) generates a prediction of whether a domain is a spam domain based on the domain features and domains with known labels. Based on the domain scores of domains, the domain ranking module may adaptively select content items from the sources with corresponding domains scores.

Abstract: A digital magazine server generates a digital magazine for user based on a received request for the digital magazine identifying one or more topics. The digital magazine server applies one or more machined trained models to obtained content items to select content items for the topic. A hierarchy of the topics included in the received request may be determined by the digital magazine server and used by the trained models to select content items. When generating the digital magazine, the digital magazine server also includes one or more editorial content items that are manually selected. The digital magazine serer may reposition one or more content items selected by the trained models to include an editorial content items.

Abstract: A digital magazine server generates a digital magazine for user based on a received request for the digital magazine identifying one or more topics. The digital magazine server applies one or more machined trained models to obtained content items to select content items for the topic. A hierarchy of the topics included in the received request may be determined by the digital magazine server and used by the trained models to select content items. When generating the digital magazine, the digital magazine server also includes one or more editorial content items that are manually selected. The digital magazine server may reposition one or more content items selected by the trained models to include an editorial content items.

Abstract: A digital magazine server generates a model to associate topics with content items. To generate the model, the digital magazine server selects a set of content items that have been included in one or more digital magazines. For each content item of the set, the digital magazine server determines a distribution of concepts associated with a content item of the set based on characteristics of digital magazines including the content item of the set and a distribution of topics associated with the content item of the set based on words included in the content item. Additionally, the digital magazine server determines a parameter defining a relationship between the distributions of concepts and the distributions of topics associated with content items of the set. A model based on the distributions of concepts and of topics as well as the parameter is generated and stored for application to content items.

Abstract: A digital magazine server displays content items from various sources to users of client devices. Each source of a content item is identified by a domain, and content items for different sources have different domain-level quality. To differentiate sources of content items, the domains identifying the sources are ranked based on domain scores of the domains generated by an aggregate of multiple trained domain classifiers. A domain score of a domain indicates a domain-level quality of content items provided by a source identified by the domain. Each of the trained domain classifiers (e.g., a naïve Bayes classifier, a random forest classifier, and a logistic regression classifier) generates a prediction of whether a domain is a spam domain based on the domain features and domains with known labels. Based on the domain scores of domains, the domain ranking module may adaptively select content items from the sources with corresponding domains scores.

Abstract: A digital magazine server displays content items from various sources to users of client devices. Each source of a content item is identified by a domain, and content items for different sources have different domain-level quality. To differentiate sources of content items, the domains identifying the sources are ranked based on domain scores of the domains generated by an aggregate of multiple trained domain classifiers. A domain score of a domain indicates a domain-level quality of content items provided by a source identified by the domain. Each of the trained domain classifiers (e.g., a naïve Bayes classifier, a random forest classifier, and a logistic regression classifier) generates a prediction of whether a domain is a spam domain based on the domain features and domains with known labels. Based on the domain scores of domains, the domain ranking module may adaptively select content items from the sources with corresponding domains scores.

Abstract: A digital magazine server generates a digital magazine for user based on a received request for the digital magazine identifying one or more topics. The digital magazine server applies one or more machined trained models to obtained content items to select content items for the topic. A hierarchy of the topics included in the received request may be determined by the digital magazine server and used by the trained models to select content items. When generating the digital magazine, the digital magazine server also includes one or more editorial content items that are manually selected. The digital magazine server may reposition one or more content items selected by the trained models to include an editorial content items.

Abstract: A digital magazine server generates a digital magazine for user based on a received request for the digital magazine identifying one or more topics. The digital magazine server applies one or more machined trained models to obtained content items to select content items for the topic. A hierarchy of the topics included in the received request may be determined by the digital magazine server and used by the trained models to select content items. When generating the digital magazine, the digital magazine server also includes one or more editorial content items that are manually selected. The digital magazine serer may reposition one or more content items selected by the trained models to include an editorial content items.

Abstract: A digital magazine server generates a personalized digital magazine tailored for the interests of a user with subtopic specificity. The digital magazine server identifies topics of interest for a user and one or more subtopics related to each topic. For each subtopic, the digital magazine server infers whether the subtopic is of interest to the user based on prior user actions performed by the user or social signals describing the user's actions with content items, such as following content items from the user's influencers. For subtopics that are of interest, relevant content items are identified and boosted in comparison to other content items that are only relevant to the topic of interest. Thus, the digital magazine server generates a personalized content feed for the user with the boosted content items, thereby enabling more relevant presentation of content items that are likely to be of interest to the user.

Abstract: A digital magazine server displays content items from various sources to users of client devices. Each source of a content item is identified by a domain, and content items for different sources have different domain-level quality. To differentiate sources of content items, the domains identifying the sources are ranked based on domain scores of the domains generated by an aggregate of multiple trained domain classifiers. A domain score of a domain indicates a domain-level quality of content items provided by a source identified by the domain. Each of the trained domain classifiers (e.g., a naïve Bayes classifier, a random forest classifier, and a logistic regression classifier) generates a prediction of whether a domain is a spam domain based on the domain features and domains with known labels. Based on the domain scores of domains, the domain ranking module may adaptively select content items from the sources with corresponding domains scores.