Home |  ENGLISH |  Kontakt |  Impressum |  Datenschutz |  Anmelden |  KIT

Thema4237

Aus Aifbportal

Wechseln zu: Navigation, Suche



Towards AI for Art and Media: Creating a Knowledge Graph for the ZKM - Zentrum für Kunst und Medien




Informationen zur Arbeit

Abschlussarbeitstyp: Master
Betreuer: Michael Färber
Forschungsgruppe: Web Science
Partner: ZKM
Archivierungsnummer: 4237
Abschlussarbeitsstatus: Offen
Beginn: 01. Mai 2020
Abgabe: unbekannt

Weitere Informationen

The ZKM would like to strengthen its capabilities in Artificial Intelligence and Digitalization. Among other things, the ZKM plans to work on chatbots, which will be available for website visitors and exhibition visitors on site. As basis for such a chatbot and potential other software systems, a knowledge graph should be developed in the pro-posed thesis. Knowledge graphs have been established as an advanced database technology and have been used by major Internet companies, such as Google and Amazon. Also in the area of art centers and museums, first institutions have created or are currently creating knowledge graphs for their needs. In this light, a knowledge graph for the ZKM should be created that will be partially publicly available and will be used as basis for various applications in the future.

The first part of the thesis will focus on the transformation of existing databases of the ZKM into a knowledge graph and to evaluate the coverage and accuracy of the knowledge graph with regard to the individual attributes. In a second step, methods should be developed and evaluated for linking phrases in a text to the ZKM knowledge graph and to freely available knowledge graphs, such as Wikidata and the German National Library catalogue. As far as time allows, first machine learning models based on the knowledge graph can be created to find similarities, differences, and analogies between artists, works of art, etc.

The thesis will be supervised by Dr. Michael Färber from KIT as well as Daniel Heiss and Margit Rosen from the ZKM.


Ausschreibung: Download (pdf)