Systems and methods for real-time keyword recommendations

A method comprising:

• receiving input information including first information about first existing content associated with a content originator, second information about keywords associated with the content originator, and third information relating to an event corpus including a future event;
• predicting based on the second information about the future event, an increase in a number of keyword searches using a keyword that will occur based on the future event;
• encoding, via a first processing pipeline that is allocated a first amount of computational resources, the first information and the third information into one or more first embeddings;
• encoding, via a second processing pipeline that is allocated a second amount of computational resources that is greater than the first amount of computational resources, the second information and the third information into one or more second embeddings;
• generating, by the first processing pipeline and second processing pipeline and using the one or more first embeddings and the one or more second embeddings, one or more pre-ranked lists of events in the event corpus to reduce latency associated with keyword associations with the content originator;
• determining, by one or more machine learning models, that content included in the one or more pre-ranked list of events is related to the keyword;
• and
• generating, in real-time or near-real-time, by the one or more machine learning models, based on the prediction, and prior to the future event, a recommendation for the content originator to perform an association between the keyword and the content such that the content or a hyperlink to the content is presented when the keyword is received in a keyword search via an application.

A system comprising:

• memory that stores computer-executable instructions;
• and
• one or more processors configured to access the memory and execute the computer-executable instructions to:
• receive input information including first information about first existing content associated with a content originator, second information about keywords associated with the content originator, and third information relating to an event corpus including a future event;
• predict based on the second information about the future event, an increase in a number of keyword searches using a keyword that will occur based on the future event;
• encode, via a first processing pipeline that is allocated a first amount of computational resources, the first information and the third information into one or more first embeddings;
• encode, via a second processing pipeline that is allocated a second amount of computational resources that is greater than the first amount of computational resources, the second information and the third information into one or more second embeddings;
• generate, by the first processing pipeline and second processing pipeline and using the one or more first embeddings and the one or more second embeddings, one or more pre-ranked lists of events in the event corpus to reduce latency associated with keyword associations with the content originator;
• determine, by one or more machine learning models, that content included in the one or more pre-ranked list of events is related to the keyword;
• and
• generate, in real-time or near-real-time, by the one or more machine learning models, based on the prediction, and prior to the future event, a recommendation to perform an association between the keyword and the content such that the content or a hyperlink to the content is presented when the keyword is received in a keyword search via an application.