Abstract
•Summarize which fuzzy techniques have emerged in big data problems.•Review when and why fuzzy techniques work in big data.•Present a critical analysis on the contributions.•Review the big data challenges and their relevance to fuzzy techniques.•Propose potential opportunities to guide further developments.
In the era of big data, we are facing with an immense volume and high velocity of data with complex structures. Data can be produced by online and offline transactions, social networks, sensors and through our daily life activities. A proper processing of big data can result in informative, intelligent and relevant decision making completed in various areas, such as medical and healthcare, business, management and government. To handle big data more efficiently, new research paradigm has been engaged but the ways of thinking about big data call for further long-term innovative pursuits. Fuzzy sets have been employed for big data processing due to their abilities to represent and quantify aspects of uncertainty. Several innovative approaches within the framework of Granular Computing have been proposed. To summarize the current contributions and present an outlook of further developments, this overview addresses three aspects: (1) We review the recent studies from two distinct views. The first point of view focuses on what types of fuzzy set techniques have been adopted. It identifies clear trends as to the usage of fuzzy sets in big data processing. Another viewpoint focuses on the explanation of the benefits of fuzzy sets in big data problems. We analyze when and why fuzzy sets work in these problems. (2) We present a critical review of the existing problems and discuss the current challenges of big data, which could be potentially and partially solved in the framework of fuzzy sets. (3) Based on some principles, we infer the possible trends of using fuzzy sets in big data processing. We stress that some more sophisticated augmentations of fuzzy sets and their integrations with other tools could offer a novel promising processing environment.