Content selection using psychological factor vectors

A system comprising:

• a server comprising one or more processors and one or more memories;
• and
• an application, executing on one or more processors of the server, configured for:
• generating a person vector, from a record, identifying a set of data representing a person;
• generating, by a neural network model, a psychological factor vector, from a content of a plurality of content, identifying a set of psychological factors from the content, wherein the psychological factor vector is generated by
• processing the content with a feature layer of the neural network model to form a plurality of feature vectors,
• processing the plurality of feature vectors with a classification layer of the neural network model to form a plurality of classification vectors;
• and
• processing the plurality of classification vectors with an aggregation layer of the neural network model to form the psychological factor vector;
• combining the person vector and the psychological factor vector to generate an input vector;
• generating an action score, from the input vector, identifying a probability of an action on the content by the person;
• selecting the content from the plurality of content using the action score;
• and
• presenting the content.

A method comprising:

• generating a training person vector, from a training record, identifying a set of data for a person;
• generating, by a neural network model, a training psychological factor vector, from a training content of a plurality of training content, identifying a set of psychological factors from the training content, wherein the training psychological factor vector is generated by
• processing the content with a feature layer of the neural network model to form a plurality of feature vectors,
• processing the plurality of feature vectors with a classification layer of the neural network model to form a plurality of classification vectors;
• and
• processing the plurality of classification vectors with an aggregation layer of the neural network model to form the training psychological factor vector;
• updating a psychological factor model using the training psychological factor vector and psychological factor feedback to train the psychological factor model;
• combining the training person vector and the training psychological factor vector to generate a training input vector;
• generating a training action score, from the training input vector, identifying a probability of an action on the training content by the person;
• and
• updating an action model using the training action score and action score feedback to train the action model.