Abstract: Certain embodiments of the present disclosure provide techniques training a user detection model to identify a user of a software application based on voice recognition. The method generally includes receiving a data set including a plurality of voice interactions with users of a software application. For each respective recording in the data set, a spectrogram representation is generated based on the respective recording. A plurality of voice recognition models are trained. Each of the plurality of voice recognition models is trained based on the spectrogram representation for each of the plurality of voice recordings in the data set. The plurality of voice recognition models are deployed to an interactive voice response system.

Abstract: Certain embodiments of the present disclosure provide techniques training a user detection model to identify a user of a software application based on voice recognition. The method generally includes receiving a data set including a plurality of voice interactions with users of a software application. For each respective recording in the data set, a spectrogram representation is generated based on the respective recording. A plurality of voice recognition models are trained. Each of the plurality of voice recognition models is trained based on the spectrogram representation for each of the plurality of voice recordings in the data set. The plurality of voice recognition models are deployed to an interactive voice response system.

Abstract: The present disclosure provides novel techniques to segment mobile entities based on signals from mobile devices associated with these mobile entities. In certain embodiments, a data file is built for each mobile entity to transform originally extremely high dimensional raw data derived from mobile devices communicating with a network into structured data characterizing associated mobile entities. The data files are structured for filtering and searching with respect to features associated with the signals from mobile devices, including location and non-location features. The data files are used to determine feature gains corresponding respectively to features related to a mobile segment. The features include first features of a first type and second features of a second type. The feature gains are used, among other things, to rank mobile entities for the mobile segment, with feature gains corresponding to the first features given a larger weight than feature gains corresponding to the second features.

Abstract: A method may include determining a combination of values of attributes represented by reference data associated with payment transaction by training a machine learning model based on an association between (i) respective values of the attributes and (ii) the payment transactions having a given result. The combination may be correlated with having the given result. The method may also include selecting a subset of the payment transactions that is associated with the combination of values. The method may additionally include determining a first rate at which payment transactions of the subset have the given result during a first time period and a second rate at which one or more payment transactions associated with the combination have the given result during a second time period, and generating an indication that the two rates differ.

Abstract: The present disclosure provides novel techniques to derive characteristic of mobile entities based signals from mobile devices associated with these mobile entities. In certain embodiments, a data file is built for each mobile entity to transform originally extremely high dimensional raw data derived from mobile devices communicating with a network into structured data characterizing associated mobile entities. The data file are structured for filtering and searching with respect to location and non-location features. A plurality of filters for a behavioral segment are applied to the data files of a large number of mobile entities. The plurality of filters including a preset hard constraint (HC) filter, a preset soft constraint (SC) filter, and an extended constraint (EC) filter built using location features in data files that have passed through the HC filter and/or the SC filter.

Abstract: The present disclosure provides novel techniques to derive characteristic of mobile entities based signals from mobile devices associated with these mobile entities. In certain embodiments, a data file is built for each mobile entity to transform originally extremely high dimensional raw data derived from mobile devices communicating with a network into structured data characterizing associated mobile entities. The data file are structured for filtering and searching with respect to location and non-location features. A plurality of filters for a behavioral segment are applied to the data files of a large number of mobile entities. The plurality of filters including a preset hard constraint (HC) filter, a preset soft constraint (SC) filter, and an extended constraint (EC) filter built using location features in data files that have passed through the HC filter and/or the SC filter.

Abstract: The disclosed technology includes system and method for impressing publicly displayed information on mobile devices. The method comprises determining at least a first place and a second place for a mobile device based on locations associated with the mobile device. Each of the first place and the second place being a place where the mobile device is regularly located as indicated by densities of the locations mapped over a geographical region and the corresponding time stamps. The first place and second place are used to determine one or more travel routes routinely taken by a user of the mobile device, and to build an association database, where the mobile device is associated with one or more information documents, each of the one or more information documents including information publicly displayed along the one or more routes.

Abstract: Certain embodiments of the present disclosure provide techniques training a user detection model to identify a user of a software application based on voice recognition. The method generally includes receiving a data set including a plurality of voice interactions with users of a software application. For each respective recording in the data set, a spectrogram representation is generated based on the respective recording. A plurality of voice recognition models are trained. Each of the plurality of voice recognition models is trained based on the spectrogram representation for each of the plurality of voice recordings in the data set. The plurality of voice recognition models are deployed to an interactive voice response system.

Abstract: Aspects of the present disclosure provide techniques for training a machine learning model. Embodiments include receiving a historical support record comprising time-stamped actions, a support initiation time, and an account indication. Embodiments include determining features of the historical support record based at least on differences between times of the time-stamped actions and the support initiation time. Embodiments include determining a label for the features based on the account indication. Embodiments include training an ensemble model, using training data comprising the features and the label, to determine an indication of an account in response to input features, wherein the ensemble model comprises a plurality of tree-based models and a ranking model.

Abstract: The disclosed technology includes system and method for impressing publicly displayed information on mobile devices. The method comprises determining at least a first place and a second place for a mobile device based on locations associated with the mobile device. Each of the first place and the second place being a place where the mobile device is regularly located as indicated by densities of the locations mapped over a geographical region and the corresponding time stamps. The first place and second place are used to determine one or more travel routes routinely taken by a user of the mobile device, and to build an association database, where the mobile device is associated with one or more information documents, each of the one or more information documents including information publicly displayed along the one or more routes.

Abstract: Location data of a large number of mobile devices and map data are used to derive a large number of commute routes in a geographical area, which are used to build a relational database including associations between a plurality of road segments and a plurality of information documents. Each of the plurality of information documents corresponds to a billboard in the geographical area and includes information displayed by the billboard. A respective road segment in the relational database has one or more associations with one or more information documents, each of the one or more information documents being associated with the respective road segment by a corresponding probability value indicating a degree of likelihood of a traveler on the respective road segment being exposed to information displayed on a billboard corresponding to the each of the one or more information documents.

Abstract: The present disclosure provides novel techniques to derive characteristic of mobile entities based signals from mobile devices associated with these mobile entities. In certain embodiments, a data file is built for each mobile entity to transform originally extremely high dimensional raw data derived from mobile devices communicating with a network into structured data characterizing associated mobile entities. The data file are structured for filtering and searching with respect to location and non-location features. A plurality of filters for a behavioral segment are applied to the data files of a large number of mobile entities. The plurality of filters including a preset hard constraint (HC) filter, a preset soft constraint (SC) filter, and an extended constraint (EC) filter built using location features in data files that have passed through the HC filter and/or the SC filter.

Abstract: The disclosed technology includes system and method for discovering key places for a mobile device based on historical location data associated with the mobile device. The key places are discovered by mapping the historical location data into predefined land areas in a geographical region, aggregating the historical location data into time interval sessions, selecting a first set of the predefined areas based on density of location data in each of the predefined land areas, and filtering the first set of the predefined areas to obtain the key places. The key places are used to estimate one or more travel routes routinely taken by a user of the mobile device, and to build an association database, where one or more billboards and/or points of interests near the travel routes are associated with the mobile device or its user. Information documents for delivering to the mobile device are selected based at least on data in the association database.

Abstract: The present disclosure provides a mobile advertising platform in which mobile user locations and other information are translated into indications of mobile user intent to approach certain businesses, and advertisers can fill mobile advertising requests or choose to price their bids for mobile supplies based on such indications. In certain embodiments, pre-defined places associated with business/brand names are created, and mobile advertising requests are processed to determine if the associated with mobile devices have triggered any of these pre-defined places. If a mobile advertising request is determined to have triggered one or more of the pre-defined places, it is annotated with the triggered place(s), and advertisements are selected based on the triggered places and other factors. The annotated requests with the triggered places can also be commodities in a location market place, which are auctioned to the mobile advertisers, who can place their bids on the triggered places.

Abstract: The present disclosure provides novel techniques to derive characteristic of mobile entities based signals from mobile devices associated with these mobile entities. In certain embodiments, a data file is built for each mobile entity to transform originally extremely high dimensional raw data derived from mobile devices communicating with a network into structured data characterizing associated mobile entities. The data file are structured for filtering and searching with respect to location and non-location features. A plurality of filters for a behavioral segment are applied to the data files of a large number of mobile entities. The plurality of filters including a preset hard constraint (HC) filter, a preset soft constraint (SC) filter, and an extended constraint (EC) filter built using location features in data files that have passed through the HC filter and/or the SC filter.

Abstract: A method according to certain embodiments comprises receiving a first request indicating a first location of a first mobile device at a first time and including first non-location data associated with the first mobile device; determining whether the first non-location data meet one or more requirements in response to the first location being proximate to a first physical object, receiving a second request indicating a second location of the first mobile device at a second time and including second non-location data; determining a set of selection factors including at least a first factor based at least on the first request in response to the second location being proximate to a second physical object; and selecting content associated with one of the first physical object and the second physical object for delivery to the mobile device based at least on the set of selection factors.

Abstract: The present disclosure provides novel techniques to derive characteristic of mobile entities based signals from mobile devices associated with these mobile entities. In certain embodiments, a data file is built for each mobile entity to transform originally extremely high dimensional raw data derived from mobile devices communicating with a network into structured data characterizing associated mobile entities. The data file are structured for filtering and searching with respect to location and non-location features. A plurality of filters for a behavioral segment are applied to the data files of a large number of mobile entities. The plurality of filters including a preset hard constraint (HC) filter, a preset soft constraint (SC) filter, and an extended constraint (EC) filter built using location features in data files that have passed through the HC filter and/or the SC filter.

Abstract: Location data of a large number of mobile devices and map data are used to derive a large number of commute routes in a geographical area, which are used to build a relational database including associations between a plurality of road segments and a plurality of information documents. Each of the plurality of information documents corresponds to a billboard in the geographical area and includes information displayed by the billboard. A respective road segment in the relational database has one or more associations with one or more information documents, each of the one or more information documents being associated with the respective road segment by a corresponding probability value indicating a degree of likelihood of a traveler on the respective road segment being exposed to information displayed on a billboard corresponding to the each of the one or more information documents.

Abstract: The disclosed technology includes system and method for discovering key places for a mobile device based on historical location data associated with the mobile device. The key places are discovered by mapping the historical location data into predefined land areas in a geographical region, aggregating the historical location data into time interval sessions, selecting a first set of the predefined areas based on density of location data in each of the predefined land areas, and filtering the first set of the predefined areas to obtain the key places. The key places are used to estimate one or more travel routes routinely taken by a user of the mobile device, and to build an association database, where one or more billboards and/or points of interests near the travel routes are associated with the mobile device or its user. Information documents for delivering to the mobile device are selected based at least on data in the association database.

Abstract: The present disclosure provides a mobile advertising platform in which mobile user locations and other information are translated into indications of mobile user intent to approach certain businesses, and advertisers can fill mobile advertising requests or choose to price their bids for mobile supplies based on such indications. In certain embodiments, pre-defined places associated with business/brand names are created, and mobile advertising requests are processed to determine if the associated with mobile devices have triggered any of these pre-defined places. If a mobile advertising request is determined to have triggered one or more of the pre-defined places, it is annotated with the triggered place(s), and advertisements are selected based on the triggered places and other factors. The annotated requests with the triggered places can also be commodities in a location market place, which are auctioned to the mobile advertisers, who can place their bids on the triggered places.