University of Twente Student Theses


Textual Clustering for Telecommunication Accident Recommendations

Malek, Michael (2022) Textual Clustering for Telecommunication Accident Recommendations.

[img] PDF
Abstract:Over the last decades, a significant amount of people became dependent on sufficiently working communication networks. A lot of these networks are proven to be fairly stable and have backup solutions, but can trigger uttermost consequences in case of an accident due to faulty machines or human error. This can lead to a tremendous number of issues for example millions of money lost or even potential death cases. Researchers worldwide are trying to improve organizations' network safety and minimize the risk factor of network downtime and accident rates. Based on another research paper concerning accident analysis in telecommunication companies by Wienen et al. \cite{16}, this paper focuses on finding insights or new patterns in textual recommendations created by experts concerning telecommunication networks. To find these new patterns, sentence embedding methods in combination with unsupervised clustering algorithms had to be applied. Sentence Transformers were applied to convert the given recommendations into a quantitative embedding matrix. 8 clustering algorithms were found and trained, of which the most optimal model was selected with internal and external validation tools and cluster visualizations. To furthermore find insightful information for the most optimal model, topic visualization tools produced word clouds that formed topics based on the most frequent words in each cluster.
Item Type:Essay (Bachelor)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Business & IT BSc (56066)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page