Examples of the disclosure provide a system and method for obtaining one or more current candidate resumes and one or more past candidate resumes associated with a role and analyzing full text of the obtained resumes to identify one or more items. Weighted values are determined for the identified items using a dimension reduction technique, and a probability score is calculated for each of the obtained current candidate resumes based on selection data associated with the obtained past candidate resumes. A keyword score is calculated for each of the obtained current candidate resumes based on a presence of one or more keywords associated with the role in the obtained current candidate resumes. A final candidate score is generated for each of the obtained current candidate resumes based on the keyword score and the calculated probability score, and the generated final candidate scores are output in association with the obtained current candidate resumes to a user interface.