<p>The adoption of heterogeneous information networks (HINs) has gained popularity as a means of modeling complex real-world systems with diverse interacting components. Despite the acknowledged advantages of HIN, a considerable number of current studies tend to treat these networks as homogeneous, neglecting the distinctions between different types of components and connections. Recently, there has been a growing recognition among researchers of the importance of treating interconnected data with multiple types as a HIN. They have developed methods to embed such data into a lower-dimensional space while preserving the diverse structures and meanings within the network. These HIN embedding methods offer versatile applications, including fraud detection, system recommendation, disease prediction, etc., leveraging the rich semantics inherent in object types and links within the network. Although HIN provides more structural and semantic information compared to homogeneous networks, it also poses challenges for data mining. In this paper, we present a comprehensive review of the approaches, techniques, and applications employed in HIN, emphasizing the diverse methodologies identified in existing literature.</p>