Unsupervised competition-based encoding

A system comprising:

• a processor;
• a memory;
• and
• the memory comprising an application, wherein the application executes on the processor, uses the memory, and is configured for:
• collecting word-based data corresponding to a first identifier;
• generating, for the first identifier, a first phrase vector by extracting frequency data from the word-based data;
• generating a similarity metric corresponding to the first identifier and a second identifier by comparing the first phrase vector of the first identifier to a second phrase vector of the second identifier;
• generating a tuple comprising the first identifier and the second identifier using the similarity metric;
• and
• training a machine learning model with the tuple to generate an embedded vector corresponding to the first identifier.

A method comprising:

• collecting word-based data corresponding to a first identifier;
• generating, for the first identifier, a first phrase vector by extracting frequency data from the word-based data;
• generating a tuple comprising the first identifier and a second identifier using the first phrase vector;
• training a machine learning model with the tuple to generate an embedded vector corresponding to the first identifier;
• clustering a set of embedded vectors to generate a cluster of embedded vectors including the embedded vector;
• adding a category corresponding to the cluster of embedded vectors to a set of categories when an additional category threshold is satisfied;
• and
• removing a previous category from the set of categories when a category removal threshold is satisfied.