Sustainability Risk Monitoring in Supply Chains - Ranking Suppliers using Text Mining and Bayesian Networks with a Focus on Child Labor

Aug 21, 2026

This thesis develops software system components for monitoring social sustainability compliance across supply chains through continuous supplier ranking by breach risk. The work concentrates on child labor and proposes a Bayesian network risk model that integrates public and private data sources. Risk is modeled with an individual Bayesian network per supplier location, while text mining extracts unstructured social sustainability incident events for BN integration. Expert evaluation and sensitivity testing validate the model, and gold-standard experiments report strong F1 scores for incident event detection (83.0%), with best performance for geography and company attributes.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages376

Reading Time15 h 48 min

AccessFree to download

Abstract

Kurzfassung

Acknowledgements

Erklärungs- und Einreichungsteil (Verfassung der Arbeit)

Sustainability Risk Monitoring in Supply Chains Ranking Suppliers using Text Mining and Bayesian Networks with a Focus on Child Labor DISSERTATION zur Erlangung des akademischen Grades Doktor der Sozial-und Wirtschaftswissenschaften eingereicht von Mag. Dipl.-Ing. Andreas Thöni, BSc Matrikelnummer 0426645 an der Fakultät für Informatik der Technischen Universität Wien Betreuung: o. Univ. Prof. Dr. A Min Tjoa o. Univ. Prof. Dr. Alfred Taudes Diese Dissertation haben begutachtet: (o. Univ. Prof. Dr. A Min Tjoa) (o. Univ. Prof. Dr. Alfred Taudes) Wien, 05.06.2015 (Andreas Thöni) Technische Universität Wien A-1040 Wien ▪ Karlsplatz 13 ▪ Tel. +43-1-58801-0 ▪ [www.tuwien.ac.at](www.tuwien.ac.at) Erklärung zur Verfassung der Arbeit Andreas Thöni Schulweg 744, A-6100 Seefeld in Tirol Hiermit erkläre ich, dass ich diese Arbeit selbständig verfasst habe, dass ich die verwendeten Quellen und Hilfsmittel vollständig angegeben habe und dass ich die Stellen der Arbeit -einschließlich Tabellen, Karten und Abbildungen -, die anderen Werken oder dem Internet im Wortlaut oder dem Sinn nach entnommen sind, auf jeden Fall unter Angabe der Quelle als Entlehnung kenntlich gemacht habe. (Ort, Datum) (Unterschrift Verfasser) Acknowledgements I am grateful to many supporters who have accompanied my work: First, to my supervisors Prof. Dr. A Min Tjoa and Prof. Dr. Alfred Taudes for their time, their encouragement, and their constant constructive feedback during the course of this work. To Dr. Jack L. King, Prof. Dr. Andreas Rauber, Dr. Alexander Schatten, and Arnavaz Schatten for the insightful discussions and suggestions. To APA, Austria Presse Agentur, for allowing me to use the APA Basisdienst Corpus. To my colleagues at university and extramural for the exchange of ideas, of motivation, and of laughters. To my dear girlfriend, Marie-Theres, for her patience and constant encouragement. Last but not least, to my parents without their support over time this thesis would not have been possible. Andreas Thöni, 1.6.2015 Vienna, Austria Abstract This thesis presents components of a software system designed to monitor social sustainability compliance in supply chains. The components allow to ongoingly rank suppliers based on their risk of breaching sustainability standards. It focuses on child labor given the extension of social sustainability and child labor’s imminent risk level. It proposes a novel risk model using a Bayesian network (BN) that allows the integration of public and private data sources. The risk level is modelled with an individual BN for each supplier location. The main focus of the work is on the model and the data integration. In order to integrate public data sources, a text mining approach is suggested to identify and extract social sustainability incident events in unstructured text. These events can be integrated into the BNs. An expert questionnaire together with sensitivity tests confirms the usefulness of the model. For the text mining task, tests on a newly developed gold standard suggest high F1 scores for incident event detection (83.0%) . Incident event extraction performs better for the geography and company event attributes and less for the sector attributes. This implies that manual input revision appears necessary, also recommended by experts interviewed. Finally, an analysis of online retrievable empirical input data reveils the availability of interesting and detailed events either published by news agencies or posted by non-governmental organizations. Kurzfassung Diese Dissertation präsentiert Komponenten für ein Softwaresystem zur Überwachung von sozialer Nachhaltigkeit in Lieferantenketten (Supply Chains) . Es ermöglicht die kontinuierliche Reihung von Lieferanten auf Basis des Risikos, dass diese soziale Nachhaltigkeitsstandards brechen. Aufgrund der Breite von sozialer Nachhaltigkeit und des imminent gegebenen Risikos geht diese Arbeit ins

FAQs

What is the main purpose of the proposed system in this thesis?

The system continuously monitors social sustainability compliance in supply chains by ranking suppliers according to the risk of breaching sustainability standards.

How is the risk level modeled for suppliers?

The thesis uses a Bayesian network, with an individual Bayesian network modeled for each supplier location to represent risk.

How are public data sources integrated into the model?

Public unstructured data are processed with a text mining approach to identify and extract social sustainability incident events, which are then integrated into the Bayesian networks.

Aug 21, 2026

Upload by Maya Linwood