Masterarbeit
Behavioral Analysis in Knowledge Graphs for Trust Awareness using ConTED
Completion
2024/12
Research Area
Students
Dipshikha Ghosh
Advisers
Description
This master's thesis outlines the evolution of web architectures, highlighting the shift from centralized to decentralized web applications. This evolution emphasizes the need for user control over personal data, advocating for a decentralized web where data is stored under user control. This research will focus on evaluating and developing the most suitable method to enhance our existing Content Trust Evaluation Framework for the Decentralized Web (ConTED) by incorporating behavioral analysis to improve trust awareness for the knowledge graphs. This involves assessing several factors like bias, deception, and incentive in content trust, aiming to provide a transparent and user-centric approach to web services.
The objective of this research is to evaluate existing algorithms or methods that can detect the values of bias, deception, and incentive in behavioral analysis, contributing to ConTED. The research aims to review relevant algorithms, compare their effectiveness, and develop or refine an algorithm suitable for these factors. The thesis will especially focus on the behavioral analysis approaches preliminarily created within recommender systems. The goal is to improve the trust awareness of a decentralized web application by creating awareness on the three ConTED factors bias, deception, and incentive within linked data resources.
The objective of this master thesis is to find an approach or combination of approaches to solve the previously mentioned problem about bias, deception, and incentive in the context of trust awareness for decentralized web applications using recommender systems. This particularly includes the state of the art regarding detection of the three factors within recommender systems. The demonstration of feasibility with an implementation demonstrator of the concept is part of this thesis as well as a suitable evaluation with exemplary use cases.