Abstract: Generating an improved/more accurate geolocation database is provided. Given a dataset of reverse DNS hostnames for IP addresses, ground truth information, and a hierarchical geographical database, a machine learning classifier can be trained to extract and disambiguate location information from the reverse DNS hostnames of IP addresses and to apply machine learning algorithms to determine location candidates and to select a most probable candidate for a reverse DNS hostname based on a confidence score. The classifier can be used to generate an accurate geolocation database, or to provide accurate geolocation information as a service.

Abstract: A user submits a query to a search system that produces a search results page having one or more segments and a plurality of URLs relevant to the query. Embodiments of the disclosure ensure cohesion between the URLs and the segments on the search results page by evaluating each segment relative to the URLs. A relevancy score is calculated for each segment based on the URLs. The relevancy score for each segment is compared to one or more thresholds. Depending on whether the relevancy score for each segment exceeds one or more thresholds, the segment can be eliminated from the search page, relocated on the search page, or left alone. Calculation of each relevancy score is based on information determined from a search log. A feedback loop can help improve the functioning of the suppression and/or relocation functions.

Abstract: A user submits a query to a search system that produces a search results page having one or more segments and a plurality of URLs relevant to the query. Embodiments of the disclosure ensure cohesion between the URLs and the segments on the search results page by evaluating each segment relative to the URLs. A relevancy score is calculated for each segment based on the URLs. The relevancy score for each segment is compared to one or more thresholds. Depending on whether the relevancy score for each segment exceeds one or more thresholds, the segment can be eliminated from the search page, relocated on the search page, or left alone. Calculation of each relevancy score is based on information determined from a search log. A feedback loop can help improve the functioning of the suppression and/or relocation functions.

Abstract: Generating an improved/more accurate geolocation database is provided. Given a dataset of reverse DNS hostnames for IP addresses, ground truth information, and a hierarchical geographical database, a machine learning classifier can be trained to extract and disambiguate location information from the reverse DNS hostnames of IP addresses and to apply machine learning algorithms to determine location candidates and to select a most probable candidate for a reverse DNS hostname based on a confidence score. The classifier can be used to generate an accurate geolocation database, or to provide accurate geolocation information as a service.

Abstract: Generating an improved/more accurate geolocation database is provided. Given a dataset of reverse DNS hostnames for IP addresses, ground truth information, and a hierarchical geographical database, a machine learning classifier can be trained to extract and disambiguate location information from the reverse DNS hostnames of IP addresses and to apply machine learning algorithms to determine location candidates and to select a most probable candidate for a reverse DNS hostname based on a confidence score. The classifier can be used to generate an accurate geolocation database, or to provide accurate geolocation information as a service.

Abstract: A method for operating a universal casting service is described. The method receives a casting request from a first client; initiates a first casting session to a first casting device; establishes a first queue for the first casting device; adds a first media item to the queue from the first client; receives a casting request from a second client; and adds a second media item to the first queue from the second client.

Abstract: A persistent viewports architecture that provides a persistent and correctly-positioned viewport to a user when a collaboration document is being updated. The architecture identifies and utilizes a user's element-of-interest in content of the document to determine and recalculate the correct position of the user viewport. The architecture tracks a topmost element of the content (or section of content) from the viewport and employs the topmost element as a reference element. The topmost element can be one or more of a paragraph, a header, string of text, and an image. When the underlying document is to be updated, the viewport performs a lookup of reference element, and after finding the reference element, adjusts the X and Y offsets of the viewport so that the element-of-interest is inside the viewport.

Abstract: Generating an improved/more accurate geolocation database is provided. Given a dataset of reverse DNS hostnames for IP addresses, ground truth information, and a hierarchical geographical database, a machine learning classifier can be trained to extract and disambiguate location information from the reverse DNS hostnames of IP addresses and to apply machine learning algorithms to determine location candidates and to select a most probable candidate for a reverse DNS hostname based on a confidence score. The classifier can be used to generate an accurate geolocation database, or to provide accurate geolocation information as a service.

Abstract: Illicit use of devices, and misappropriation of the data on those devices, may be prevented by detecting patterns of behavior that suggest illicit use, and by taking action when such illicit use is detected. A device may store information that describes suspicious patterns of use, and may also store information that describes normal patterns of use of known legitimate users. If current use of the device matches a suspicious pattern of use, and if the user of the device cannot be confirmed to be a known legitimate user, then action may be taken, such as locking the device's user interface, shutting down the device's communication facilities, issuing a warning message, sending a communication, or any other action.

Abstract: A method for operating a universal casting service is described. The method receives a casting request from a first client; initiates a first casting session to a first casting device; establishes a first queue for the first casting device; adds a first media item to the queue from the first client; receives a casting request from a second client; and adds a second media item to the first queue from the second client.

Abstract: Illicit use of devices, and misappropriation of the data on those devices, may be prevented by detecting patterns of behavior that suggest illicit use, and by taking action when such illicit use is detected. A device may store information that describes suspicious patterns of use, and may also store information that describes normal patterns of use of known legitimate users. If current use of the device matches a suspicious pattern of use, and if the user of the device cannot be confirmed to be a known legitimate user, then action may be taken, such as locking the device's user interface, shutting down the device's communication facilities, issuing a warning message, sending a communication, or any other action.

Abstract: A persistent viewports architecture that provides a persistent and correctly-positioned viewport to a user when a collaboration document is being updated. The architecture identifies and utilizes a user's element-of-interest in content of the document to determine and recalculate the correct position of the user viewport. The architecture tracks a topmost element of the content (or section of content) from the viewport and employs the topmost element as a reference element. The topmost element can be one or more of a paragraph, a header, string of text, and an image. When the underlying document is to be updated, the viewport performs a lookup of reference element, and after finding the reference element, adjusts the X and Y offsets of the viewport so that the element-of-interest is inside the viewport.

Abstract: Systems, methods, and computer-readable media are presented regarding a viewport adjustment module that provides content-aware scrolling. A viewport adjustment module receives notice of scrolling by a viewport. A determination is made as to the end position of the viewport where the scrolling will stop. A determination is made as to whether new content displayable in the viewport at the end position will be only partially displayable in the viewport. A determination is made regarding an adjustment amount to the end position of the viewport such that the new content in the viewport is fully displayable in the viewport. The end position of the viewport is adjusted according to the determined adjustment amount.