The paper introduces a novel technique for multilabel classification and ranking by extending label ranking methods to incorporate a calibrated scenario. Label ranking typically operates on an uncalibrated scale, lacking a natural zero point, which restricts its applicability to certain tasks like multilabel classification. The proposed method introduces an artificial calibration label that separates relevant from irrelevant labels, enhancing the expressive power of existing label ranking approaches. This technique combines pairwise preference learning and relevance classification, where a separate classifier predicts the relevance of each label. Empirical results on text categorization, image classification, and gene analysis datasets demonstrate the effectiveness of the calibrated model compared to state-of-the-art multilabel learning methods. The approach is shown to improve both the selection of relevant labels and ranking performance, making it a valuable addition to the field of multilabel learning.The paper introduces a novel technique for multilabel classification and ranking by extending label ranking methods to incorporate a calibrated scenario. Label ranking typically operates on an uncalibrated scale, lacking a natural zero point, which restricts its applicability to certain tasks like multilabel classification. The proposed method introduces an artificial calibration label that separates relevant from irrelevant labels, enhancing the expressive power of existing label ranking approaches. This technique combines pairwise preference learning and relevance classification, where a separate classifier predicts the relevance of each label. Empirical results on text categorization, image classification, and gene analysis datasets demonstrate the effectiveness of the calibrated model compared to state-of-the-art multilabel learning methods. The approach is shown to improve both the selection of relevant labels and ranking performance, making it a valuable addition to the field of multilabel learning.