Abstract: Systems and methods are described that include determining an intent from an intent library that matches a received query, identifying, based on the intent, at least one mini-app from a first provider and at least one mini-app from a second provider, ranking the identified mini-apps, and triggering rendering of search results, the search results including at least a first ranked mini-app in an expanded state and the remaining ranked mini-apps in a collapsed state.

Abstract: Systems and methods are described that include determining an intent from an intent library that matches a received query, identifying, based on the intent, at least one mini-app from a first provider and at least one mini-app from a second provider, ranking the identified mini-apps, and triggering rendering of search results, the search results including at least a first ranked mini-app in an expanded state and the remaining ranked mini-apps in a collapsed state.

Abstract: Systems and methods are described that include determining an intent from an intent library that matches a received query, identifying, based on the intent, at least one mini-app from a first provider and at least one mini-app from a second provider, ranking the identified mini-apps, and triggering rendering of search results, the search results including at least a first ranked mini-app in an expanded state and the remaining ranked mini-apps in a collapsed state.

Abstract: Systems and methods are described that include determining an intent from an intent library that matches a received query, identifying, based on the intent, at least one mini-app from a first provider and at least one mini-app from a second provider, ranking the identified mini-apps, and triggering rendering of search results, the search results including at least a first ranked mini-app in an expanded state and the remaining ranked mini-apps in a collapsed state.

Abstract: Systems and methods inject related content to a search result using a dwell score. For example, a method may include, for at least one result at least partially displayed in a viewport, calculating a dwell score for the result at based on an amount of time the result has been in the viewport and a position of the result in the viewport and, if the dwell score meets a threshold, displaying suggested links for the result in the viewport. Another example method includes displaying a scrollable set of search results in a viewport and, while waiting for a scroll action or link selection, calculating a dwell score for at least one result that has content visible in the viewport and updating ranking signals based on the dwell score. The updated ranking signals are used to generate a next set of search results for the query.

Abstract: Implementations are described for determining, based on a user-assembled collection of electronic content items, at least one actual purpose for the assembled collection. Those implementations are further directed to selecting an additional electronic content item based on the at least one actual purpose being assigned to the additional electronic content item in an electronic database. Those implementations are yet further directed to, responsive to the selecting, causing a suggestion for the additional electronic content item to be rendered at a client device associated with a user account that is assigned to the user-assembled collection.

Abstract: Systems and methods are described that include determining an intent from an intent library that matches a received query, identifying, based on the intent, at least one mini-app from a first provider and at least one mini-app from a second provider, ranking the identified mini-apps, and triggering rendering of search results, the search results including at least a first ranked mini-app in an expanded state and the remaining ranked mini-apps in a collapsed state.

Abstract: Systems and methods inject related content to a search result using a dwell score. For example, a method may include, for at least one result at least partially displayed in a viewport, calculating a dwell score for the result at based on an amount of time the result has been in the viewport and a position of the result in the viewport and, if the dwell score meets a threshold, displaying suggested links for the result in the viewport. Another example method includes displaying a scrollable set of search results in a viewport and, while waiting for a scroll action or link selection, calculating a dwell score for at least one result that has content visible in the viewport and updating ranking signals based on the dwell score. The updated ranking signals are used to generate a next set of search results for the query.

Abstract: Implementations are described for determining, based on a user-assembled collection of electronic content items, at least one actual purpose for the assembled collection. Those implementations are further directed to selecting an additional electronic content item based on the at least one actual purpose being assigned to the additional electronic content item in an electronic database. Those implementations are yet further directed to, responsive to the selecting, causing a suggestion for the additional electronic content item to be rendered at a client device associated with a user account that is assigned to the user-assembled collection.

Abstract: Implementations are described for determining, based on a user-assembled collection of electronic content items, at least one actual purpose for the assembled collection. Those implementations are further directed to selecting an additional electronic content item based on the at least one actual purpose being assigned to the additional electronic content item in an electronic database. Those implementations are yet further directed to, responsive to the selecting, causing a suggestion for the additional electronic content item to be rendered at a client device associated with a user account that is assigned to the user-assembled collection.