An automated resume ranking system selects and sorts relevant resumes from those sent in response to a job offer (JO). During the screening and elimination process, resume content is largely analyzed, while JO details are only marginally considered. In this sense, existing resume ranking approaches lack the accuracy necessary to detect relevant information in JOs, which is imperative to ensure that selected resumes are relevant to the JO. This study examines the uncertainty-based estimation to assess 16 textual markers applied to extract relevant terms in JOs—10 textual markers obtained by examining the behavior of expert recruiters and 6 from the literature—based on two approaches: fuzzy logistic regression and fuzzy decision trees. Results indicate that, globally, fuzzy decision trees improve the F1 and recall metrics by 27% and 53% respectively, compared to state-of-the-art term extraction techniques.