Community detection, an NP-hard problem, aims to partition networks into substructures to reveal their latent functions. As social network scales expand, traditional community detection algorithms face challenges in accuracy and efficiency, and computational resource consumption has increased dramatically. The application of computational intelligence has balanced the effectiveness and time cost of community detection algorithms. This paper presents a classification framework for community detection algorithms based on computational intelligence, divided into three main categories: evolutionary computation-based, neural networks-based and fuzzy theory-based. It further discusses the core ideas and unique solutions of these methods, guiding researchers in selecting appropriate algorithms. Additionally, the paper explores current challenges and future research directions, aiming to advance community detection technology in handling complex social network data.