Strongly connected component (SCC) detection is fundamental for analyzing citation graphs, yet existing general-purpose algorithms inefficiently handle the dynamic nature and specific properties of these networks. This study addresses this gap by developing specialized incremental SCC detection methods. We first leverage distinct edge types inherent in citation graphs to devise partition and local topological ordering strategies, minimizing redundant graph traversals. Based on this, we introduce two efficient bounded incremental algorithms: one for continuous single updates via dynamic maintenance of partitions and order, and the other for batch updates that further reduces edge traversals by building upon the single-update technique. Experimental evaluations on real-world citation graphs verify significant efficiency improvements, with our single incremental method achieving speedups of at least 11.5 times, and the batch incremental method achieving speedups of at least 5.0 times compared with baseline methods.

$K$-nearest neighbor ( $K$NN) is one of the most fundamental methods for unsupervised outlier detection because of its various advantages, e.g., ease of use and relatively high accuracy. Currently, most data analytic tasks need to deal with high-dimensional data, and the $K$NN-based methods often fail due to “the curse of dimensionality”. AutoEncoder-based methods have recently been introduced to use reconstruction errors for outlier detection on high-dimensional data, but the direct use of AutoEncoder typically does not preserve the data proximity relationships well for outlier detection. In this study, we propose to combine $K$NN with AutoEncoder for outlier detection. First, we propose the Nearest Neighbor AutoEncoder (NNAE) by persevering the original data proximity in a much lower dimension that is more suitable for performing $K$NN. Second, we propose the $K$-nearest reconstruction neighbors ( $K$NRNs) by incorporating the reconstruction errors of NNAE with the $K$-distances of $K$NN to detect outliers. Third, we develop a method to automatically choose better parameters for optimizing the structure of NNAE. Finally, using five real-world datasets, we experimentally show that our proposed approach NNAE+ $K$NRN is much better than existing methods, i.e., $K$NN, Isolation Forest, a traditional AutoEncoder using reconstruction errors (AutoEncoder-RE), and Robust AutoEncoder.