The present work was submitted to the Institute of Information Systems and Databases RWTH Aachen University Faculty 1 - Mathematics, Computer Science and Natural Sciences Course of Studies - Computer Science Master’s thesis An Efficient Semantic Search Engine for Research Data in an RDF-based Knowledge Graph communicated by Prof. Dr. Stefan Decker Examiners: Author: Prof. Dr. Stefan Decker Sarah Bensberg, 378908 Prof. Dr. Matthias Müller Dr. Marius Politze Aachen, October 16, 2020 Statutory Declaration in Lieu of an Oath Bensberg, Sarah 378908 Last Name, First Name Matriculation No. (optional) I hereby declare in lieu of an oath that I have completed the present Master thesis entitled An Efficient Semantic Search Engine for Research Data in an RDF-based Knowledge Graph independently and without illegitimate assistance from third parties (such as academic ghostwriters). I have used no other than the specified sources and aids. In case that the thesis is additionally submitted in an electronic format, I declare that the written and electronic versions are fully identical. The thesis has not been submitted to any examina- tion body in this, or similar, form. Aachen , October 16, 2020 City, Date Signature Official Notification: Para. 156 StGB (German Criminal Code): False Statutory Declarations Whoever before a public authority competent to administer statutory declarations falsely makes such a declaration or falsely testifies while referring to such a declaration shall be liable to impris- onment not exceeding three years or a fine. Para. 161 StGB (German Criminal Code): False Statutory Declarations Due to Negligence (1) If a person commits one of the offences listed in sections 154 through 156 negligently the penalty shall be imprisonment not exceeding one year or a fine. (2) The offender shall be exempt from liability if he or she corrects their false testimony in time. The provisions of section 158 (2) and (3) shall apply accordingly. I have read and understood the above official notification: Aachen , October 16, 2020 City, Date Signature Acknowledgements At this point, I would like to thank everyone who supported me during the preparation of this master thesis. First of all, I would like to thank my supervisor Dr. Marius Politze for giving me the opportunity to conduct my master thesis in the Process and Application Development Research group at the IT Center of RWTH Aachen University, where the infrastructure used in the thesis was provided. I am grateful for the challenging tasks, much support, and helpful advice. My thanks go to Prof. Dr. Stefan Decker and Prof. Dr. Matthias Müller, who kindly agreed to supervise and review my work. I would like to thank them very much for their helpful suggestions and constructive criticism during the creation of this work. Furthermore, I deeply thank all members of the department for their interest in my work, their patience, and their helpfulness. Finally, I thank my family and my friends for supporting me, keeping me motivated, and always being there for me, not only during the time of my master thesis. Abstract An Efficient Semantic Search Engine for Research Data in an RDF-based Knowledge Graph Sarah Bensberg Digital transformation affects all areas of society: More data is produced and workflows rely on data analysis leading to new challenges in data management. Within the context of research data management, the National Research Data Infrastructure aims to system- atize these data stocks and make them accessible. RWTH Aachen University supports this effort with the development of the research data management platform CoScInE. Re- search data is made accessible independent of the actual storage location and described with metadata that allows a structured search. The goal is to make the data findable, accessible, interoperable, and reusable, according to the FAIR Guiding Principles. The W3C standards for the semantic web, such as the data model RDF, the corresponding query language SPARQL, and related technologies, provide the means to describe digital resources but require a significant amount of technical knowledge. This thesis deals with the implementation of a research data search in an RDF knowledge graph that is less dependent on the users’ background in knowledge engineering. Different approaches are considered under several evaluation criteria and it is investigated whether mapping the RDF data into a search index for use in a search engine improves the quality of the search and results. Such an implementation is opposed to the systematic generation of a SPARQL query. In the course of the thesis, a transformation of RDF graphs into a search index using application profiles and rules was developed. This allows the use of all functionalities and search syntaxes provided by the search engine applied. The evaluation shows that in most cases an approach is either easy to use but slow or ineffective, or, on the contrary, fast and effective but difficult to use. With the presented transformation, a solution was found that combines these two contradictory properties. Zusammenfassung Eine effiziente semantische Suchmaschine für Forschungsdaten in einem RDF-basierten Wissensgraphen Sarah Bensberg Der digitale Wandel wirkt sich auf alle Bereiche der Gesellschaft aus: Es werden stetig mehr Daten produziert und Arbeitsprozesse sind auf Datenanalysen angewiesen. Dies führt zu neuen Herausforderungen im Datenmanagement. Im Rahmen des Forschungsdatenman- agements zielt die Nationale Forschungsdateninfrastruktur darauf ab, diese Datenbestände zu systematisieren und zugänglich zu machen. Die RWTH Aachen University unter- stützt dieses Vorhaben mit der Entwicklung der Forschungsdatenmanagement Plattform CoScInE. Forschungsdaten werden unabhängig von ihrem eigentlichen Speicherort zugäng- lich gemacht und mit Metadaten beschrieben, die eine strukturierte Suche ermöglichen. Ziel ist es die Daten nach den FAIR Leitprinzipien auffindbar, zugänglich, interoperabel und wiederverwendbar zu machen. Die W3C-Standards für das semantische Web, wie das Datenmodell RDF, die zugehörige Abfragesprache SPARQL und verwandte Technologien, bieten die Mittel zur Beschreibung digitaler Ressourcen, erfordern jedoch ein erhebliches Maß an technischem Wissen. Diese Arbeit befasst sich mit der Implementierung einer Forschungsdatensuche in einem RDF- Wissensgraphen, die unabhängig vom fachlichen Hintergrund des Nutzers ist. Dazu werden verschiedene Ansätze unter mehreren Evaluationskriterien betrachtet und der Hypothese nachgegangen, ob die Abbildung der RDF Daten in einen Suchindex zur Verwendung in einer Suchmaschine die Qualität der Suche und Ergebnisse verbessern. Eine solche Implementierung steht der systematischen Generierung einer SPARQL Abfrage gegenüber. Im Rahmen der Arbeit wurde eine Transformation von RDF Graphen in einen Suchindex mithilfe von Applikationsprofilen und Regeln entwickelt. Hierdurch können alle durch die verwendete Suchmaschine bereitgestellten Funktionalitäten und Suchsyntaxen verwendet werden. Die Evaluation zeigt, dass ein Ansatz in den meisten Fällen entweder einfach zu bedienen, jedoch langsam oder ineffektiv ist oder im Gegenteil schnell und effektiv, dafür aber schwierig zu benutzen ist. Mit der vorgestellten Transformation wurde ein Lösung gefunden, welche diese beiden konträren Eigenschaften vereint. Contents 1 Introduction1 1.1 NFDI.......................................1 1.2 Motivation....................................1 1.3 RDM at RWTH Aachen University.......................2 1.4 CoScInE......................................3 1.5 Research Goals and Questions..........................4 1.6 Methodology and Structure...........................6 2 Fundamentals9 2.1 Metadata and Related Concepts........................9 2.2 Semantic Web and Linked Data......................... 10 2.2.1 RDF Data Model............................. 10 2.2.2 RDF Vocabularies............................ 11 2.2.3 Validate and Constrict RDF Data using SHACL and DASH..... 12 2.2.4 Query and Update RDF Data using SPARQL............. 13 2.2.5 Infer over RDF Data........................... 14 2.2.6 Open and Closed World Assumption.................. 14 2.3 Information Retrieval.............................. 16 2.3.1 Unstructured, Structured, and Semi-Structured Data......... 16 2.3.2 Search Queries.............................. 16 2.3.3 Different Approaches of Information Retrieval............. 16 2.4 Semantic Search................................. 17 2.4.1 Entity Retrieval Model for Web Data................. 17 2.4.2 Entity Attribute-Value Model...................... 18 2.4.3 Search Model for Web Data....................... 18 2.5 Related Work................................... 19 3 Technical Details of CoScInE 25 3.1 Data Model.................................... 25 3.2 Database Model................................. 29 3.3 Components and Processes........................... 29 3.4 Search Requirements............................... 30 3.4.1 General Conditions............................ 30 3.4.2 Effectiveness............................... 31 3.4.3 Usability................................. 31 3.4.4 Complexity of Search Request for the User.............. 31 3.4.5 Efficiency................................. 31 3.4.6 Response Time.............................. 31 3.4.7 Scalability................................. 32 3.4.8 Additional Effort............................. 32 4 Different Approaches 33 4.1 Literal and Additional Rules........................... 33 4.2 Full-Text Search in RDF Literals........................ 37 4.3 SPARQL Query Builder............................