System and method for validating a candidate recommendation model

The computer implemented method of clam 3, further comprising:

• identifying a set of training parameters from a plurality of training parameters in response to determining that the ML ranking model is not valid;
• training the ML ranking model based on the identified set of training parameters;
• causing the synthetic test data to be applied to the trained ML ranking model, the trained ML ranking model being programmed to generate new prediction data based on the synthetic test data;
• and
• evaluating the new prediction data relative to the predicted performance data to determine whether the trained ML ranking model is valid.

A system comprising:

• memory to store machine readable instructions, a text dataset comprising a plurality of different types of text data characterizing different texts, and metadata that includes parameter logic for populating at least a work experience parameter for a plurality of applicants and candidates and a job requisition parameter for a job requisition;
• one or more processors configured to access the memory and execute the machine readable instructions, the machine readable instructions comprising:
• a synthetic data generator programmed to:
• populate the work experience parameter with a given portion of text of respective text data from the text dataset based on the parameter logic, wherein the work experience parameter is computed for each of the plurality of applicants or candidates based on different text data from the dataset;
• populate the job requisition parameter with a different portion of text of the respective text data used to compute the first parameter for a given applicant or candidate of the plurality of applicants or candidates;
• and
• generate synthetic test data based on the computed work experience parameter and job requisition parameter to test a learning to rank (LTOR) model trained on training data that is from a different data source than the text dataset to validate a performance of the LTOR model;
• and
• a performance evaluator programmed to:
• apply the synthetic test data to the LTOR model to cause the LTOR model to generate a ranking list to provide a ranking of each of the plurality of applicants or candidates relative to the job requisition;
• and
• evaluate the generated ranking list relative to predicted performance data characterizing a predicted ranking performance of the LTOR model to determine whether the LTOR model is valid.

One or more non-transitory computer readable medium having instructions executable by a processor to perform a method, the method comprising:

• applying a weighting algorithm to respective text data from a text dataset comprising a plurality of different types of text data to identify a given portion of text of the respective text data;
• generating a first parameter for each of a plurality of applicants or candidates based on corresponding text data from the text dataset, the first parameter generated for a given applicant or candidate of the plurality of applicants and candidates being populated with the given portion of the text of the respective text data, and all first parameters for remaining applicants or candidates is populated with a respective portion of text from different text data as used to populate the first parameter for the given applicant or candidate;
• applying the weighting algorithm to the respective text data from the text dataset to identify a different portion of text of the respective text data;
• generating a second parameter with the different portion of the text of the respective text data for the given applicant or candidate of the plurality of applicants and candidates;
• and
• generating synthetic test data based on the computed first and second parameters to test a machine learning (ML) ranking model trained on training data that is from a different data source than the text dataset to validate a performance of the ML ranking model.