While a wealth of statutory and auditing pronouncements attest to the importance of the auditing of journal entries for preventing and detecting material misstatements to financial statements, existing literature has so far paid inadequate attention to this line of research. To explore this line of research further, this paper proposes a bipartite model that is based on extreme value theory and Bayesian analysis of Poisson distributions. The paper assesses the veracity of the model via a series of experiments on a dataset that contains the journal entries of an international shipping company for fiscal years 2006 and 2007. Empirical results suggest the model can detect journal entries that have a low probability of occurring and a monetary amount large enough to cause financial statements to be materially misstated. Further investigations reveal that the model can assist auditors to form expectations about the journal entries thus detected as well as update their expectations based on new data. The findings indicate that the model can be applied for the auditing of journal entries, and thus supplement existing procedures.

A considerable body of regulatory pronouncements attests to the significance of auditing journal entries for ensuring that financial statements are free of material misstatements; however, existing empirical studies have paid insufficient attention to the audit of journal entries. To explore this issue further, this paper proposes a model based on self-organizing map as well as validates this model by performing experiments on a dataset containing journal entries. Empirical results suggest that the proposed model can detect "suspicious" and legitimate transactions with a high degree of accuracy. Further investigations reveal that the performance of the model is robust to varying prior probabilities of "suspicious" journal entries occurring in the population. The findings indicate that the model can assist auditors in detecting "suspicious" journal entries.

The self-organizing map (i.e. SOM) has inspired a voluminous body of literature in a number of diverse research domains. We present a synthesis of the pertinent literature as well as demonstrate, via a case study, how SOM can be applied in clustering accounting databases. The synthesis explicates SOM's theoretical foundations, presents metrics for evaluating its performance, explains the main extensions of SOM, and discusses its main financial applications. The case study illustrates how SOM can identify interesting and meaningful clusters that may exist in accounting databases. The paper extends the relevant literature in that it synthesises and clarifies the salient features of a research area that intersects the domains of SOM, data mining, and accounting. 

A considerable body of literature attests to the significance of internal controls; however, little is known on how the clustering of accounting databases can function as an internal control procedure. To explore this issue further, this paper puts forward a semi-supervised tool that is based on self-organizing map and the IASB XBRL Taxonomy. The paper validates the proposed tool via a series of experiments on an accounting database provided by a shipping company. Empirical results suggest the tool can cluster accounting databases in homogeneous and well-separated clusters that can be interpreted within an accounting context. Further investigations reveal that the tool can compress a large number of similar transactions, and also provide information comparable to that of financial statements. The findings demonstrate that the tool can be applied to verify the processing of accounting transactions as well as to assess the accuracy of financial statements, and thus supplement internal controls.

The application of Self-Organizing Map (SOM) to hierarchical data remains an open issue, because such data lack inherent quantitative information. Past studies have suggested binary encoding and Generalizing SOM as techniques that transform hierarchical data into numerical attributes. Based on graph theory, this paper puts forward a novel approach that processes hierarchical data into a numerical representation for SOM-based clustering. The paper validates the proposed graph-theoretical approach via complexity theory and experiments on real-life data. The results suggest that the graph-theoretical approach has lower algorithmic complexity than Generalizing SOM, and can yield SOM having significantly higher cluster validity than binary encoding does. Thus, the graph-theoretical approach can form a data-preprocessing step that extends SOM to the domain of hierarchical data.