Semantic services as a symbol-based artificial intelligence technology can be used in various application scenarios, e.g., enhancing deep-learning systems, autonomous driving, or knowledge-based systems. With SPARQL, there is a widely accepted query language standard available, allowing uniform access regardless of the underlying technology - similar to the SQL-standard for relational databases. But while the landscape of relational storage is highly mature and almost all SQL-based storage systems have high performances, this is not the case for triple-based databases. These, often research-driven developments, are more diverse regarding their functionalities and efficiency. This makes the selection of the right tool for the selected goals complex.
Your task is the evaluation of free-to-use RDF-based triple-stores like apache fuseki or virtuoso. Using uniform queries, you will measure the performance of the software and compare the expected results. Based on your findings, you can derive recommendations which database software is best to use for which application case.

Knowledge graphs are increasingly being used to make use of the large number of different data sources in companies. These link a large number of different data on the basis of ontologies and allow implicit knowledge to be converted into explicit knowledge. They are hereby the cornerstone of rule-based artificial intelligence.
In research, the Department of Information Systems at the University of Rostock successfully cooperates with industry to ensure the maximum benefit of ontologies and knowledge graphs. In particular, evaluation and quality assurance based on structural properties are research foci. However, since the underlying information is usually always business-critical and accordingly strictly confidential, sharing the data is often a problem. The ontologies are often not allowed to leave the corporate boundaries.
The thesis advertised here is intended to remedy this: In the bachelor thesis, an anonymization method shall be developed. This is to anonymize the underlying data so far that all structural characteristics remain, but the business-critical data are removed as far as possible. The terms are to be replaced by other contents (e.g. from Wordnet). The process shall be reversible by dropping the given transformations. The implementation shall be done as a Python script.

Rule-based artificial intelligence is increasingly being used in enterprises in the form of knowledge graphs to break down the often isolated, siloed data structures. Here, the ontologies provide the underlying set of rules: they contain the knowledge of how the various pieces of information are related. These are mostly defined in the RDF(S) and OWL languages. However, the focus of these languages is primarily on knowledge reasoning, not data modeling. This makes them partly unsuitable for mapping business rules. This led to the development and standardization of the Shape Constraint Language (SHACL).
The University of Rostock offers a tool for data-driven, automatic evaluation of RDF(S) and OWL based ontologies: NEOntometrics (neontometrics.com). The software computes ontology metrics based on properties of, for example, the graph, annotations, or type of logical statements. However, the increased use of SHACL for modeling knowledge creates a gap in the evaluation, as SHACL is not yet covered by NEOntometrics.
The goal of this work is to develop an automatic evaluation of SHACL in ontologies. For this purpose, first the current state of knowledge in the literature will be explored. Then you will create a software concept and implement this in Python. The evaluation of the research will be based on freely available ontologies that contain SHACL.