Kategorie: ‘Theses’
Algorithmic Approaches to Overlapping Community Detection – Label Propagation Algorithm with Neighbor Node Influence
A Practical Look at Membership Inference Attacks and Differential Privacy Protection on Data with Re-occurring Information
Decentralized Identity and Access Management for Distributed Machine Learning Systems
Bridging the gap between design and deployment of statistical analyses in Distributed Analytics
Development of a Data Ecosystem Model for the Context of Procurement
Algorithmic Approaches to Overlapping Community Detection – Local Optimization Algorithm based on Cliques
Algorithmic Approaches to Overlapping Community Detection – Clustering Graphs into Subgraphs
Overlapping Community Detection for Botnet analysis on microblogging plattforms
The study of (social) structures in real-world networks is known as (social) network analysis. Nodes and linkages make up networks. Communities are subnetworks in which nodes have more connections to nodes within the subnetwork than to nodes outside the subnetwork. The difficulty of detecting nodes in networks that belong to more than one sub-network is known as overlapping community detection. The goal of this work is to extend our existing bot detection framework to include community detection functionality.
Time-series based Academic Trend and Downtrend Detection
Analysis of literature in scientific conferences and journals to identify and describe the evolution of trends is valuable for supporting the dissemination and distribution of knowledge in a large-scale decentralized research community. A trend is an increase in popularity, while a downtrend describes a decrease in popularity of a certain topic. The goal of this work is to extend our existing citation recommendation bot to provide temporal context information to the recommendations by time-series based trend analysis.