summaery2020: Projects

Voice Search with Knowledge Graphs

Project information

submitted by
Johannes Kiesel

Co-Authors
Ronald Kurniawan, Josef Roth

Mentors
Marcel Gohsen, Johannes Kiesel, Benno Stein

Faculty:
Media

Degree programme:
Computer Science and Media (Bachelor of Science (B.Sc.)),
Computer Science and Media (english) (Master of Science (M.Sc.))

Type of project presentation
Research project

Semester
Summer semester 2021


Project description

We use a question-answering (QA) system as a guide in a virtual-reality replication of the Gropiuszimmer. The Gropiuszimmer serves here as a room to illustrate the history of the Weimar Bauhaus. In this project, we use a combination of a question-answering and a voice-interaction system as the guide to the virtual room. We acquire the information about Gropiuszimmer from a human expert and model it as a graph structure.

QA systems give an appropriate answer to a question expressed in a natural language (e.g. English). To be able to do that, the question has to be parsed into parts understandable by the QA system. The parts are then formed into a SPARQL query, which gives the appropriate information in our knowledge graph. Generally, QA systems perform great with general information (e.g. the capital of a country). To use information specific to the Gropiuszimmer, we also incorporate knowledge about it into our knowledge graph. We use a web survey positioned in the virtual room to gain possible topics, questions, and thoughts of users about objects, persons, facts, and stories that belong to the Gropiuszimmer. We do so by simulating the tour guide providing prompts to the study participants.

Email: johannes.kiesel[at]uni-weimar.de

Exhibition Location / Event Location

  • – ONLINE –
  • – ONLINE –
  • – ONLINE –
  • – ONLINE –