This work addresses the problem of synthesizing fuzzy temporal logic rules from a set of given positive and negative examples. The examples are provided in the form of execution traces of finite length. Fuzzy Time Linear Temporal Logic over finite traces ( \(FTL_f\) ) is chosen as the language for rule synthesis. \(FTL_f\) is capable of capturing fuzzy temporal modalities, like, ‘soon after’, ‘almost always’, ‘gradually’ etc., that make the learnt rules simpler and more understandable than classical LTL representations. The proposed approach reduces the learning task to a multi-valued partial maximum satisfiability (PMaxSAT) problem. This work is useful for generating interpretable explanations of complex system behaviours.