Multi-stage content analysis system that profiles users and selects promotions

A multi-stage content analysis system that profiles users and selects promotions, comprising:

• a computer comprising a memory and instructions within said memory, wherein said computer is configured to execute said instructions to
• receive a communication created by a user via a message analyzer;
• analyze said communication, via said message analyzer, to determine whether said user is a potential target for one or more promotions, wherein said analyze said communication comprises
• access a database comprising promotion information that comprises information that indicates how to determine which promotion of said one or more promotions is an appropriate response to said communication,
• scan the communication and match said communication against said promotion information in said database using one or more of text analysis, natural language processing, keyword matching or artificial intelligence,
• wherein said determine whether said user is said potential target for said one or more promotions depends on a type of said one or more promotions,
• wherein when said communication matches said promotion information, a determination is made by said computer that said user is said potential target for said one or more promotions;
• based on and in response to said analyze said communication, when said computer determines that said user is said potential target for a promotion
• receive a communications history associated with said user via a communication history analyzer, wherein said communications history comprises a plurality of communications created by said user;
• and,
• when said computer receives said communications history based on when said computer determines that said user is said potential target for said one or more promotions, analyze said communications history, via said communication history analyzer, to confirm or reject said determination that said user is said potential target for said one or more promotions, wherein said confirm or reject comprises
• assign one or more user profile tags to said user, wherein said one or more user profile tags describe characteristics of one or more of said user and said communications history of said user, wherein said analyze said communications history to assign said one or more user profile tags to said user comprises
• parse said communications history using a natural language processing classifier to derive features associated with said characteristics,
• wherein said features comprise n-grams, words or phrases across different messages within said communications history,
• assign different weights to said features, such that said features are weighted differently depending on one or more of
• a position of a message associated with said features within said communications history across said different messages, and
• a location of said features within said message;
• and,
• receive said one or more user profile tags;
• based on and in response to said analyze said communications history and in response to said receive said one or more user profile tags,
• analyze said one or more user profile tags and said communication to determine whether a specific promotion of said one or more promotions should be provided to the user, via a promotion selector, based on whether said one or more user profile tags are relevant to any promotions of said one or more promotions,
• wherein when the one or more user profile tags do not contain tags relevant to any promotions, via said promotion selector,
• no promotion is selected, or
• a random promotion is selected to assist with machine learning,
• wherein when the one or more user profile tags do contain tags relevant to said any promotions, via said promotion selector,
• select said specific promotion from said one or more promotions;
• transmit said specific promotion to said user;
• and,
• track
• said specific promotion,
• said communication, and
• statistics describing one or more of
• whether said user responded to said specific promotion that was transmitted to said user;
• how said user responded to said specific promotion;
• purchases, subscriptions, or enrollments made by said user;
• a response of said user to said specific promotion;
• incorporate said specific promotion, said communication and said data into a response record;
• and,
• assemble said response record into a training dataset;
• a machine learning engine configured to
• receive said response record including said training dataset;
• execute a machine learning algorithm on said training dataset to
• train said machine learning engine by learning or refining models that are applied, and
• update one or more of said message analyzer, said communication history analyzer and said promotion selector,
• wherein the training dataset indicates which of said one or more promotions are associated with positive responses from said user having certain user profile tags of said one or more user profile tags, to improve promotion selection in future communications.