A systematic literature review was conducted to examine anomaly detection methods. A paper was written on the results in which techniques and trends are reviewed and research gaps are identified. The paper was accepted at the “2022 3rd International Conference on Pattern Recognition and Machine Learning (PRML 2022)” and will be presented from July 15-17 and subsequently published (link to paper follows). Attendance at the conference will be virtual due to the current situation.
To enable the explanation of anomalies in a hypothesis-free testing approach, several methods have been developed and evaluated that are suitable for the explanation of anomalies on mixed-type data. The developed approaches as well as the evaluation are presented in a paper which has now been accepted at the “34th International Conference on Software Engineering & Knowledge Engineering”. The paper will be presented there from July 01 to 10 and will be published afterwards (link to the paper follows). Due to the current situation, SEKE22 will take place virtually.
In this study, the approach of “document retrieval” was taken up in order to provide auditors with the possibility to automatically retrieve relevant regulatory documents based on search queries. The results were presented in a paper entitled “Query-Based Retrieval of German Regulatory Documents for Internal Auditing Purposes”, which was accepted at the 5th International Conference on Data Science and Information Technology (DSIT 2022) and will be presented and subsequently published from July 22 to 24.
On December 16, 2021, the second year-end meeting was held with the project managers, the steering committee and the project leaders. The main content of the meeting included the presentation of the work packages and milestones already completed. As part of this presentation, a live demo of the prototype was performed as well. Subsequently, the next steps for the last project year were presented and discussed. At the end of the meeting, the already completed as well as the planned scientific publications were presented and the feedback of all participants was summarized.
From the middle of June Dennis S. will support us with 10 hours a week in the development of our backend! Dennis is a master student in the study program Business Informatics and will work on the analysis of document layouts – especially on the identification and extraction of table contents. We are looking forward to work with Dennis in the future and welcome him to our team!
As part of a comprehensive literature review, the use of Natural Language Processing in internal auditing was investigated. A paper was written about the results, which was accepted at the “Americas Conference on Information Systems (AMCIS) 2021” and will be presented between August 9 to 13 and subsequently published (link to the paper follows). AMCIS 2021 will be held virtually due to the current situation.
The first year-end meeting with the project managers, the steering committee and the project leaders took place on December 10, 2020. The main content of the meeting included the presentation of the work packages and milestones already completed. As part of this presentation, a live demo of the current status of the prototype was performed, among other things. Subsequently, the next steps planned for the short, medium and long term were presented. As a conclusion of the year-end meeting, the already completed as well as planned scientific publications were presented and the feedback of all participants was summarized.
In the DIfA project, the use of anomaly detection using autoencoders was evaluated in the context of audit tests. The results were published in a paper titled “Using Autoencoders for Data-Driven Analysis in Internal Auditing”, which was accepted at the Hawaii International Conference on System Sciences (HICSS) 2021 and will be presented and published from January 04 to 09. HICSS 2021 will be held virtually and participation is free of charge for visitors (https://hicss.hawaii.edu/)
In mid-September 2020, two master’s theses started, whose topics are embedded in the DIfA project. Analogous to the structure of the DIfA project, the focus of these master theses is on the areas of unsupervised anomaly detection and natural language processing. The two master students are supervised by the project leaders Jakob Nonnenmacher and Gerrit Schumann.