Improving Creativity Queries using Clustering Algorithms


  • proficient in Java 
  • basic knowledge of data structures and algorithms (clustering algorithms and foundations of AI course taken an optional plus)
  • English communication and writing skills - B2
Academic Advisor
Cognitive Systems, Artificial Intelligence, Computational Creativity, Creative Cognition
Bachelor of Science (B.Sc.) OR Master of Science (M. Sc.) - depth of the project will differ at the different levels


Project context

The Creative Cognitive Systems (CreaCogs) project studies creativity and creative problem solving in natural and artificial cognitive systems -

The aims of this project are twofold. One is the study of creative problem solving in humans. The second is the implementation of, study and experimentation with artificial cognitive systems which yield similar performance as human participants and can be evaluated with human creativity tests. These systems are built on the bases of cognitive knowledge acquisition and cognitively inspired knowledge organization and processes. The Creative Cognitive Systems (CreaCogs) project is supported by the German Research Foundation (DFG).

Problem statement

Creativity is sometimes assessed as a function of the ability to come up with associations. For example, in the Remote Associates Test  (Mednick, 1971), a participant is given three query words, and asked to come up with a fourth, that is associated to each of them. For example, the following three items could be given: COTTAGE, SWISS and CAKE. A good answer could be CHEESE, which associates with each of these three items, because of the concepts COTTAGE CHEESE, SWISS CHEESE and CHEESECAKE. 

As part of the CreaCogs project, an ample set of creativity queries has been constructed for the Remote Associates Test (Olteteanu et al, 2017) and showed to allow new and stronger empirical designs (Olteteanu and Schultheis, 2017). However, examining the underpinnings of creativity query formation has revealed multiple gaps in the theory of creativity. For example, what exactly is meant by "remoteness", when dealing with such associates? In order to understand this, remoteness needs to be explored in a computationally informed manner, taking into account cognitive factors of item association. 


To use various clustering algorithms on associates and determine various classes of remoteness in RAT queries. To explore the influence this remoteness has on the quality of creativity queries.

To provide a computationally informed definition of item remoteness in creativity queries. 


  • Mednick, S. A., & Mednick, M. (1971). Remote associates test: Examiner’s manual. Houghton Mifflin.
  • Oltețeanu, Ana-Maria; Schultheis, Holger and Dyer, Jonathan B. (2017) - Computationally constructing a repository of compound Remote Associates Test items in American English with comRAT-G, in: Behavior Research Methods, doi:10.3758/s13428-017-0965-8.
  • Oltețeanu, Ana-Maria and Schultheis, Holger (2017) - What determines creative association? Revealing two factors which separately influence the creative process when solving the Remote Associates Test, in: The Journal of Creative Behaviour, in press, doi:10.1002/jocb.177.