Abstract

Fraud on the event logs derived from business process is known as event log-based fraud. Fraud detection using event logs employs process mining technique, notably conformance checking analysis. This study proposes a method for rating event log-based fraud datasets using fuzzy analytic hierarchy process (AHP), a well-known multicriteria decision-making method. Earlier studies proposed that interval type-2 fuzzy set provides an alternative in handling uncertainty than type-1 fuzzy set. Therefore, we utilize interval type-2 fuzzy sets in fuzzy AHP in order to manage vagueness in many linguistic judgement. This study includes linguistic hedges implementation to modify the membership function of expert valuation. The experimental results showed comparable types of membership function shape and its obtained accuracy performance. Obtained accuracy from fuzzy type-1 AHP method achieved 94% for triangular membership function shape and 93.9% for interval type-2 fuzzy AHP. Although there was no escalation in accuracy after applying interval-valued fuzzy sets for all scenarios, the rank of fraud weight in each feature were altered.

Original languageEnglish
Title of host publicationProceedings of the 2016 IEEE Region 10 Conference, TENCON 2016
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1965-1968
Number of pages4
ISBN (Electronic)9781509025961
DOIs
Publication statusPublished - 8 Feb 2017
Event2016 IEEE Region 10 Conference, TENCON 2016 - Singapore, Singapore
Duration: 22 Nov 201625 Nov 2016

Publication series

NameIEEE Region 10 Annual International Conference, Proceedings/TENCON
ISSN (Print)2159-3442
ISSN (Electronic)2159-3450

Conference

Conference2016 IEEE Region 10 Conference, TENCON 2016
Country/TerritorySingapore
CitySingapore
Period22/11/1625/11/16

Keywords

  • even-log based fraud
  • fraud rating
  • fuzzy AHP
  • interval type-2 fuzzy sets

Fingerprint

Dive into the research topics of 'Event log-based fraud rating using interval type-2 fuzzy sets in fuzzy AHP'. Together they form a unique fingerprint.

Cite this