Method and apparatus for providing personalized self-help experience

A non-transitory computer-readable storage medium containing instructions that, when executed, cause one or more processors to perform operations for personalizing a self-support service in an online application coupled with an online community forum, the operations comprising:

• training a predictive model through a machine training process to learn a relationship between one or more user features and one or more tags based on historical user feature data;
• extracting a plurality of tags from the online community forum containing a plurality of question-answer pairs;
• creating high-dimensional representations of each tag based on a contextual relationship between each tag and other words in the online community forum by using a neural network model;
• mapping each respective question of the plurality of question-answer pairs to a set of tags;
• forming high-dimensional vectors for each respective question of the plurality of question-answer pairs by averaging the high-dimensional representation of the set of tags into a final high-dimensional vector;
• obtaining user data when an online user takes actions within the online application;
• predicting, by the predictive model, tag probabilities based on the user data for the online user by running a predictive model;
• putting weights on respective high-dimensional representations of each tag of the tag probabilities by respective tag probabilities;
• averaging the weighted high-dimensional vectors of tags of the tag probabilities into a user high-dimensional vector;
• calculating a plurality of similarity metrics, wherein each respective similarity metric measures the similarity between the high-dimensional vector for a respective question and the user high-dimension vector;
• and
• providing one or more of the plurality of question-answer pairs to the online user based on the plurality of similarity metrics.

A system for personalizing a self-support service in an online application coupled with an online community forum, comprising:

• a transmitter for transmitting data and a receiver for receiving data;
• and
• a processor configured to:
• train a predictive model through a machine training process to learn a relationship between one or more user features and one or more tags based on historical user feature data;
• extract a plurality of tags from the online community forum containing a plurality of question-answer pairs;
• create high-dimensional representation of each tag based on a contextual relationship of each tag within the online community forum by using a neural network model;
• map each respective question of the plurality of question-answer pairs to a set of tags;
• form high-dimensional vectors for each respective question of the plurality of question-answer pairs based on the set of tags mapped to the respective question;
• cause the receiver to obtain user data when a user take actions within an online application;
• predict, by the predictive model, tag probabilities for the online user based on the user data;
• put weights on respective high-dimensional representations of each tag of the tag probabilities by respective tag probabilities;
• average the weighted high-dimensional vectors of tags of the tag probabilities into a user high-dimensional vector;
• calculate a plurality of similarity metrics, wherein each respective similarity metric measures the similarity between each high-dimensional vector for a respective question and the user high-dimensional vector;
• and
• provide one or more of the plurality of question-answer pairs to the online user based on the plurality of similarity metrics.