Abstract: Methods and systems are provided for processing tag-based event communications using machine learning. One or more event communications are received from a user device. The communication(s) include key-value pairs representing an ordered sequence of multiple interaction events of a set of predefined events. Each communication of the one or more event communications includes one generated via execution of tag code integrated with code of an app page or of a webpage. A representation of the ordered sequence is processed using a machine learning model to generate one or more profile estimation results that include an identification of a particular user profile from amongst a set of stored user profile. Profile data is transmitted to a client system that identifies the particular user profile or is from the particular user profile.

Abstract: The current document is directed to automated market-segment-discovery methods and systems that may be incorporated within, or used in combination with, various types of analysis and optimization automated systems for automated discovery of market segments for subsequent use in targeted marketing and information distribution. In one implementation, a log of visitor records collected by an analysis and/or optimization system is processed to generate a segment-discovery tree. Construction of the segment-discovery tree produces a set of candidate market-segment-defining rules. Various different techniques and metrics can be applied to produce a set of market-segment-defining rules from these candidate rules. The market-segment-defining rules can then be exported to marketing systems or subsystems to facilitate targeted marketing and information distribution.

Abstract: The current document is directed to methods and systems for testing web sites. In certain implementations of the methods and systems, a testing service collects customer page-access and conversion information on behalf of a web site. The testing service is straightforwardly accessed and configured, through a web-site-based user interface, and is virtually incorporated into the web site by simple HTML-file modifications. A more efficient web-site-testing system nonuniformly distributes web-site accesses among web-page variants in order to more quickly and computationally efficiently determine a most effective web-page variant among a set of tested web-page variants. In certain implementations, nonuniform distribution of web-site accesses among web-page variants is facilitated by a Bayesian-inference method.

Abstract: 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.

Abstract: 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 a topic or category to each of a number of events. In a described implementation, the events comprise key/value pairs. A seeded local/global-topic latent Dirichlet allocation methods is used to discover topics and assign topics to a set of events. The topic-assigned events are then processed to generate topic signatures, using which the methods and systems assign topics to subsequently received messages.

Abstract: Methods and systems are provided for processing tag-based event communications using machine learning. One or more event communications are received from a user device. The communication(s) include key-value pairs representing an ordered sequence of multiple interaction events of a set of predefined events. Each communication of the one or more event communications includes one generated via execution of tag code integrated with code of an app page or of a webpage. A representation of the ordered sequence is processed using a machine learning model to generate one or more profile estimation results that include an identification of a particular user profile from amongst a set of stored user profile. Profile data is transmitted to a client system that identifies the particular user profile or is from the particular user profile.

Abstract: 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.

Abstract: The current document is directed to methods and systems for testing web sites. In certain implementations of the methods and systems, a testing service collects customer page-access and conversion information on behalf of a web site. The testing service is straightforwardly accessed and configured, through a web-site-based user interface, and is virtually incorporated into the web site by simple HTML-file modifications. A more efficient web-site-testing system nonuniformly distributes web-site accesses among web-page variants in order to more quickly and computationally efficiently determine a most effective web-page variant among a set of tested web-page variants. In certain implementations, nonuniform distribution of web-site accesses among web-page variants is facilitated by a Bayesian-inference method.

Abstract: 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 a topic or category to each of a number of events. In a described implementation, the events comprise key/value pairs. A seeded local/global-topic latent Dirichlet allocation methods is used to discover topics and assign topics to a set of events. The topic-assigned events are then processed to generate topic signatures, using which the methods and systems assign topics to subsequently received messages.

Abstract: The current document is directed to automated market-segment-discovery methods and systems that may be incorporated within, or used in combination with, various types of analysis and optimization automated systems for automated discovery of market segments for subsequent use in targeted marketing and information distribution. In one implementation, a log of visitor records collected by an analysis and/or optimization system is processed to generate a segment-discovery tree. Construction of the segment-discovery tree produces a set of candidate market-segment-defining rules. Various different techniques and metrics can be applied to produce a set of market-segment-defining rules from these candidate rules. The market-segment-defining rules can then be exported to marketing systems or subsystems to facilitate targeted marketing and information distribution.