In data science, there are still a variety of uncertainty attributes of objects which can’t be accurately represented numerically, under which case, fuzzy mathematics provides technologies and theoretical guarantee for the representations of their uncertain attributes. So far, fuzzy representation and fuzzy set theory have been widely used in decision support systems and fuzzy reasoning, and fuzzy graph is an effective approach for the structured representation of fuzzy data. The coloring of fuzzy graph plays an irreplaceable role in task allocation and conflict coordination. This work summarizes various existing definitions of fuzzy coloring, analyzes the extant fuzzy fractional coloring which is currently proposed, and gives a new definition of fuzzy fractional coloring in terms of a different version of fuzzy coloring.