Principal Investigator:

Federal Ministry of Education and Research (BMBF) [project number 03IO1617]

Nov 01, 2016 — Mar 31, 2020


One of the main processes in innovation is the generation of ideas for new applications of technologies. This process can be implemented using collaborative ideation platforms (i.e. online large-scale platforms for generating ideas), which has shown remarkable success so far: Leveraging the crowd allows the generation of a large number of ideas and the heterogeneity of the crowd increases the potential for high diversity of ideas due to participants' different backgrounds. However, finding valuable ideas has proven challenging: Related work on open innovation challenges found out that having the users generate ideas without some form of inspiration leads to mundane and repetitive ideas. Furthermore the large amount of ideas makes it unfeasible to check every idea manually.

At the Human-Centered-Computing lab, we research how to address these challenges by leveraging semantic technologies and human-computer collaboration. Our approach is based on three key steps:

Understanding users' ideas

We obtain information about the users' ideas, by extracting entities from the submitted idea text and link them to existing knowledge bases (e.g. Wikidata and Dbpedia, using SPARQL) [1]. By using an interactive validation approach, we get insights about the ideas without distracting the ideators from their main task of generating ideas.

Improving the ideation outcome for each individual user

Related work has shown that providing exemplars (inspiring ideas of others) greatly improves the idea generation outcome in terms of diversity and novelty of the generated ideas. We carry out research on constructing a semantic model (i.e. extracting super-classes, calculating conceptual similarity, finding relationships between ideas) and using it to chose effective inspirations for the ideators [2].

Improving the understandability of idea generation results

The availability of semantic knowledge about the ideas will enable us to present the outcome of an idea generation result in various visualizations. For example, calculating the conceptual similarity between ideas allows us to build a two-dimensional representation of all ideas, the so-called 'solution map'. This solution map can be used to get a quick overview over group efforts and to detect idea clusters.

By building a semantic model of the ideas generated, we are also taking additional steps towards our greater vision: Building an interactive system that, based on an idea-based knowledge graph, actively collaborates with the user to generate new and valuable ideas (so-called Human-Computer Co-Creativity). This visionary system could provide automatic recombination of ideas, adaptive system behavior based on the users mental state or customized inspiration provision based on idea relationships.

[1]: Khiat, A.; Mackeprang, M.; Müller-Birn, C. (2017): OntoIdea: Ontology-based Approach for Enhancing Collaborative Ideation. In Proceedings of Semantics Conference, Amsterdam.

[2]: Mackeprang, M.; Khiat, A.; Müller-Birn, C. (2018): Concept Validation during Collaborative Ideation and Its Effect on Ideation Outcome. In Proceedings of the 2018 CHI Conference Extended Abstracts on Human Factors in Computing Systems, Montréal.