Monitoring Database Use Patterns for Anomaly Detection: A Report

This report provides an overview of database anomaly detection, exploring various patterns and techniques used to identify unusual data behavior. It discusses the challenges in anomaly detection, including the need for sufficient labeled information. The report categorizes anomalies into update, deletion, and insert types and highlights the use of conditional and marginal patterns for detecting individual record anomalies. It also covers time-series anomaly detection, network intrusion detection, and the application of Bayesian networks and machine learning approaches, such as density-based and support vector machine-based methods. The report references key research papers on the topic, providing a comprehensive analysis of database anomaly detection strategies and their practical implications.

Monitoring database use patterns to detect anomalies

In order to detect the anomalies in databases, the patterns are used for providing the mechanisms
which allow the data administrators for granting the application programs. This discussion will
include the patterns which are used for detecting the anomaly which is a challenge in databases.
Anomaly is considered a pattern of observations that do not show the normal behavior of the
data. It requires the appropriate actions for the timely detections.
Anomalies detection includes the techniques for identifying the unusual patterns which do not
confirm the behavior. The main challenge in the detection of anomalies in the database includes
obtaining the enough label information for characterizing anomalies. The databases majorly
include the collection for which are set of attributes.
Anomalies are of three types which are update, deletion and insert anomalies. The patterns are a
conditional and marginal method which helps in detecting individual record anomalies and
ignore the rare values and the type of database are the categorical values.
According to Djenouri (2019), Anomaly pattern detection is also included which records the
groups of records with the low self-similarity in the groups. The other pattern is anomalous
group detection which helps in recording the high self-similarity in the groups but the anomalous
score is low. In monitoring databases, time-series anomaly detection includes the automatic
surveillance system which helps in monitoring the time series data for detecting the
abnormalities. In databases, network intrusion detection is also included which uses the
techniques of a survey that uses the entropy for capturing the unusual changes for inducing the
anomalies in distributing the traffic.
According to Akoglu, et al., 2012, the detection of anomalies in databases can be done by using
the Bayesian network which helps in presenting the probability models for the attributes by
categorizing the data by using the parameters for the inference techniques and efficient learning.
A typical anomaly detection method is learning Bayesian networks use training data to calculate
the likelihood of each record in a test Give a dataset of Bayesian network models and report test
records as potential anomalies with extremely low probability
During the monitoring of the database, the usage of patterns is required for detecting anomalies
for the resources and the performance of the database is done by creating and maintaining high
performance. The anomalies profile algorithm is used for creating the runtime behavior profile

for the written applications. In this, the net is used for detecting the anomalies which include
malicious emails and disease outbreak detection (Mathew, et al., 2010).
A statistical method is used in the database for identifying the data from the deviations. In the
case of a machine learning approach, the database uses the pattern for detecting anomalies
includes the density bases which includes the clustering-based and support vector-machine based
on detecting the anomalies which are typically associated with the extensions (Vela, et al.,
2017).

Reference
Akoglu, L., Tong, H., Vreeken, J., & Faloutsos, C. (2012, October). Fast and reliable anomaly
detection in categorical data. In Proceedings of the 21st ACM international conference on
Information and knowledge management (pp. 415-424).
Djenouri, Y., Belhadi, A., Lin, J. C. W., Djenouri, D., & Cano, A. (2019). A survey on urban
traffic anomalies detection algorithms. IEEE Access, 7, 12192-12205.
Mathew, S., Petropoulos, M., Ngo, H. Q., & Upadhyaya, S. (2010, September). A data-centric
approach to insider attack detection in database systems. In International Workshop on
Recent Advances in Intrusion Detection (pp. 382-401). Springer, Berlin, Heidelberg.
Vela, A. P., Ruiz, M., & Velasco, L. (2017). Distributing data analytics for efficient multiple
traffic anomalies detection. Computer Communications, 107, 1-12.