Method and system for cold start candidate recommendation

A system for candidate recommendation comprising:

• memory to store machine readable instructions and data, the data comprising resume data, a resume index, a baseline resume machine learning (ML) model trained based on non-industry-relevant resume training data, and a candidate re-ranking parameter;
• and
• one or more processors to access the memory and execute the machine readable instructions, the machine readable instructions comprising:
• a search request interface programmed to receive a query request for a resume list comprising at least a job search parameter that includes job information data and job title information for a job;
• a coarse search query parser programmed to execute a first search query, the first search querying comprising:
• generating a job search parameter vector to represent the job search parameter and a resume vector, the resume vector being generated by applying each resume of the resume data to the baseline resume ML model;
• comparing the job search parameter vector and each resume vector to identify a subset of resumes from a plurality of resumes of the resume data;
• and
• ranking the subset of resumes based on assigned scores for the subset of resumes to provide an initial ranked resume list;
• a re-ranking parser selector programmed to select a cold start query parser from a set of re-ranking parsers comprising the cold start query parser and a learning to rank (LTOR) query parser;
• wherein the cold start query parser is programmed to execute a second search query, the second search query comprising:
• evaluating the resume index to identify a set of resumes from the subset of resumes based on the job search parameter;
• and
• re-ranking the set of resumes according to a ranking function and based on the candidate re-ranking parameter to provide an updated ranked resumes list corresponding to the resume list;
• and
• an ML model generator programmed to track a training progression of a resume ML model corresponding to a ranking quality of the resume ML model based on industry-relevant resume training data,
• wherein the re-ranking parser selector is further programmed to select the LTOR query parser from the set of re-ranking parsers based on the training progression of the resume ML model indicating that the resume ML model provides recommendations at a greater ranking quality than the baseline resume ML model, and
• wherein the LTOR query parser is programmed to execute a third search query based on the resume ML model for another query request for resume ranking.