Community detection in networks is a key topic in modern network science. Communities, or clusters, are groups of vertices more likely to connect to each other than to other groups. However, defining communities is ambiguous, leading to confusion and misconceptions. This paper provides a guide to the main aspects of community detection, discusses strengths and weaknesses of popular methods, and offers guidance on their use. The paper begins with an introduction to network science, explaining that networks consist of vertices and edges. Communities are dense subgraphs that are well-separated from each other. However, communities can also overlap, and their definition varies. The paper discusses the classic view of communities as dense subgraphs and the modern view, which considers probabilistic relationships between vertices. The paper then addresses the validation of community detection methods. It discusses artificial benchmarks, partition similarity measures, detectability, and the role of metadata. It emphasizes the importance of artificial benchmarks in evaluating algorithms and the need for careful consideration of partition similarity measures. The paper then presents various methods for community detection, including consensus clustering, spectral methods, overlapping communities, and dynamic clustering. It also discusses the significance of detected communities and the choice of method. The paper concludes with a discussion of software tools for community detection and an outlook on future research. It emphasizes the importance of understanding the limitations of community detection methods and the need for careful validation of results. The paper also highlights the importance of considering the structure and metadata of networks when evaluating community detection methods.Community detection in networks is a key topic in modern network science. Communities, or clusters, are groups of vertices more likely to connect to each other than to other groups. However, defining communities is ambiguous, leading to confusion and misconceptions. This paper provides a guide to the main aspects of community detection, discusses strengths and weaknesses of popular methods, and offers guidance on their use. The paper begins with an introduction to network science, explaining that networks consist of vertices and edges. Communities are dense subgraphs that are well-separated from each other. However, communities can also overlap, and their definition varies. The paper discusses the classic view of communities as dense subgraphs and the modern view, which considers probabilistic relationships between vertices. The paper then addresses the validation of community detection methods. It discusses artificial benchmarks, partition similarity measures, detectability, and the role of metadata. It emphasizes the importance of artificial benchmarks in evaluating algorithms and the need for careful consideration of partition similarity measures. The paper then presents various methods for community detection, including consensus clustering, spectral methods, overlapping communities, and dynamic clustering. It also discusses the significance of detected communities and the choice of method. The paper concludes with a discussion of software tools for community detection and an outlook on future research. It emphasizes the importance of understanding the limitations of community detection methods and the need for careful validation of results. The paper also highlights the importance of considering the structure and metadata of networks when evaluating community detection methods.