Skip to content. | Skip to navigation

Informatik 5
Information Systems
Prof. Dr. M. Jarke
Sections
Personal tools
You are here: Home Staff Prof. Dr. Stefan Decker

Prof. Dr. Stefan Decker - Theses


Bachelor
Developing a Data Annotation Tool for Scientific Data Management - Open
Supervised by Prof. Dr. Stefan Decker; Advisor(s): Dr. Oya Deniz Beyan
Semantic technologies and RDF data representation can improve the reusability of scientific data and enable scientist to reproduce the experiments. However there is no tool to support researcher for making their data semantically interpretable by computers. Aim of the thesis is to understand the benefits of semantic web technologies for reproducible research and develop a tool which can convert experimental data to RDF by annotating with selected data models.
Master
Exploring Unknown Environments - Finding Pollution in Underground Pipes - Open
Supervised by Prof. Dr.-Ing. Gerd Ascheid, Prof. Dr. Stefan Decker; Advisor(s): Dr. Michael Cochez, Ahmed Hallawa
Bachelor
Dynamic Embeddings of Evolving Knowledge Graphs - Running
Supervised by Prof. Dr. Stefan Decker; Advisor(s): Dr. Michael Cochez, Dr. Florian Lemmerich
The goal of this Bachelor thesis is the research of updating KG embeddings with new information in order to obtain a dynamic and stable embedding of the fast-evolving KG while reducing the computational effort.
Master
Graph-Structured Query Construction for Natural Language Questions - Running
Supervised by Prof. Dr. Stefan Decker; Advisor(s): Dr. Michael Cochez
Graph-structured queries provide an efficient means to retrieve desired data from large-scale knowledge graphs. However, it is difficult for non-expert users to write such queries, and users prefer expressing their query intention through natural language questions. Recently, an increasing effort is being exerted to construct graph-structured queries for given natural language questions. At the core of the construction is to deduce the structure of the target query and retrieve vertices/edges of the underlying knowledge graph which constitute the query. Existing query construction methods rely on conventional graph-based algorithms and question understanding techniques, which lead to inefficient and degraded performances facing complicated natural language questions over knowledge graphs with large scales. In this thesis, we focus on this problem and propose novel construction models standing on recent knowledge graph embedding techniques. Extensive experiments were conducted on question answering benchmark datasets, and the results demonstrate that our models outperform baselines in terms of effectiveness and efficiency.
Bachelor
Machine Learning for Anonymization of Unstructured Text - Running
Supervised by Prof. Dr. Stefan Decker; Advisor(s): Dr. Michael Cochez
This thesis addresses the problem of identifying personal information in unstructured text using supervised Machine Learning (ML). The final application should be able to recognize and annotate the tokens that make up personal data in an English input text as accurately as possible. First, supervised learning methods, suitable for the task, will be identified. Then, models based on the most promising approaches will be designed and implemented. For comparison, suitable evaluation metrics have to be determined. Finally, the approaches are compared and evaluated against a baseline and each other.
Master
Modeling for Street Level Crime Prediction - Running
Supervised by Prof. Dr. Stefan Decker; Advisor(s): Dr. Michael Cochez, Cristina Kadar, Raquel Rosés Brüngger
The aim of this master thesis is to build a predictive model of crime at street level for a Swiss city, including a tool implementation for visualizing the data and results.
Master
Patterns for Integrating Rule Based and Process Based Model Components of Computerized Clinical Guidelines - Running
Supervised by Prof. Dr. Stefan Decker, Dr. rer. nat. Cord Spreckelsen; Advisor(s): 692050c6199c8bbfb9be2189e82ff904
Master
Privacy Attack on Social Networks Using Network Embeddings - Running
Supervised by Prof. Dr. Markus Strohmaier, Prof. Dr. Stefan Decker; Advisor(s): Dr. Florian Lemmerich, Dr. Michael Cochez
Abstract. A company that runs a social network trains a node embedding on the network where each account is represented by one node. One user deletes his account. Thus, the company is legally required to remove all private information of that user. This includes the node associated with the user’s account and the vector representation of that node that is generated by the embedding. The company, however, does likely not delete the vector representations of the other nodes even though the removed node was used during training of these. Is it possible to identify the neighbors of the removed node? Which kinds of neighbors can be identified best, which cannot be identified? First results suggest that the identification of neighbors works well for some kind of nodes and is more difficult for others.
Master
Semantic Anomaly Detection in Medical Time Series - Running
Supervised by Prof. Dr. Stefan Decker, Dr. rer. nat. Cord Spreckelsen; Advisor(s): Dr. Oya Deniz Beyan
Master
Unsupervised Anomaly Detection in Medical Time Series Data - Finished
Completed by Brockhoff, Tobias in 2019; Supervised by Prof. Dr. Stefan Decker; Advisor(s): Dr. Oya Deniz Beyan
Bachelor
Concept embeddings for Wikipedia across language editions - Finished
Completed by Felix Ingenerf in 2019; Supervised by Prof. Dr. Markus Strohmaier, Prof. Dr. Stefan Decker; Advisor(s): Dr. Florian Lemmerich, Dr. Michael Cochez
Wikipedia is a free and openly available source of information curated by users. The content available varies between language versions. The question is now whether the content available, and specifically the associations between articles, is dependent or influenced by cultural differences between users (readers and editors) in different parts of the world. In this thesis the student investigates whether these could be found trough graph embeddings which are created on the Wikipedia link graph, the graph formed by interactions with Wikipedia and a graph formed by measuring similarity between pages.