In: Genetic and Evolutionary Computing, Proceedings of the Twelfth International Conference on Genetic and Evolutionary Computing, pp.383--391.
ISSN/ISBN: Not available at this time. DOI: 10.1007/978-981-13-5841-8_40
Abstract: In forensic analysis, such as forensic auditing, multimedia forensic, and financial fraud detection, the auditor needs to detect data tempering to find clue for possible fraud. First digit distribution such as Benford’s law is proved to be an efficient tool and is used by many auditing companies to preprocess the data before the actual auditing. However, when the range of the data is limited, the first digit distribution usually does not conform to Benford’s law. Using temperature data from a sensor network, we show that if the data can be modeled by a graph signal model, then after the graph Fourier transformation, the distribution of first digits conforms to a generalized Benford’s law. In addition, a graphic model based on historical data provides better fit to the Benford’s model than that based on geodesic distance. This model is evaluated for simulated data and temperature sensor network. This finding may help to build models for forensic analysis of accounting data and sensor network data for fraud detection.
Bibtex:
@incollection{,
author = {Shi, Li-Fang and Yan, Bin and Pan, Jeng-Shyang and Sun, Xiao-Hong},
title = {Generalized Benford's Distribution for Data Defined on Irregular Grid},
year = {2019},
pages = {383--391},
booktitle = {Genetic and Evolutionary Computing},
isbn = {978-981-13-5840-1},
doi = {10.1007/978-981-13-5841-8_40}
}
Reference Type: Conference Paper
Subject Area(s): Accounting