Motifs are regarded as network blocks because motifs can be used to present fundamental patterns in networks. Motif discovery is well applied in various scientific problems, including subgraph mining and graph isomorphism tasks. This paper analyzes and summarizes current motif discovery algorithms in the field of network science with both efficiency and accuracy perspectives. In this paper, we present motif discovery algorithms, including MFinder, FanMod, Grochow, MODA, Kavosh, G-tries, QuateXelero, color-coding approaches, and GPU-based approaches. Based on that, we discuss the real-world applications of the algorithms mentioned above under different scenarios. Since motif discovery algorithms are diffusely demanded in many applications, several challenges may be firstly handled, including high computational complexity, higher order motif discovery, same motif detection, discovering heterogeneous sizes of motifs, as well as motif discovery results visualization. This work sheds light on current research progress and future research orientations.

主题被视为网络块，因为主题可用于呈现网络中的基本模式。主题发现已很好地应用于各种科学问题，包括子图挖掘和图同构任务。本文从效率和准确性的角度分析和总结了网络科学领域当前的主题发现算法。在本文中，我们介绍了主题发现算法，包括MFinder，FanMod，Grochow，MODA，Kavosh，G-tries，QuateXelero，颜色编码方法和基于GPU的方法。基于此，我们讨论了上述算法在不同情况下的实际应用。由于图案发现算法在许多应用中被广泛使用，因此可能首先要应对一些挑战，包括高计算复杂性，高阶主题发现，相同主题检测，发现主题大小不一以及可视化主题发现结果。这项工作揭示了当前的研究进展和未来的研究方向。