Summary of all notations in this paper.

The experimental results of IISOM and traditional SOM on our own test dataset.

Qualitatively judgments are given from very bad (− −) over average (o) to very good (++).

Sample Twitter popular keywords and tweets associated with normal social dynamics.

(PDF)

A broad spectrum of size, dimensionality and anomaly percentage is covered. They also differ in difficulty and cover local and global anomaly detection tasks.

The metrics results of IISOM, traditional SOM and SOMSA on KDD cup dataset.

The comparisons of TSOM, SOMSA and IISOM on the detection time.

The models used include VARMA and SVR, whereas the features used include the raw social dynamics (Baseline), level-1 activation vectors (H1), level-2 activation vectors (H2),...

The last row shows a comparison with the best nearest-neighbor method for the dataset.

This data set contains the data collected on the DAVIDE HPC system (CINECA & E4 & University of Bologna, Bologna, Italy) in the period March-May 2018. The data set has...

Four different strategies for keeping the components have been used for rPCA, while for HBOS the number of different bins was altered.

The comparisons of TSOM, SOMSA and IISOM on detection accuracy.

The impact of map size of SOM on the quality of IISOM.

The models used include VARMA and SVR, whereas the features used include the raw social dynamics (Baseline), level-1 activation vectors (H1), level-2 activation vectors (H2),...

The experimental results of IISOM and traditional SOM on our own test dataset.

The keywords are underlined.

The metrics results of IISOM, traditional SOM and SOMSA on Real dataset.