The research of the GENIUS Lab is intended to lead to the production of more reliable and accurate knowledge representations, which can inform and guide intelligent and sustained decision-making processes. The ultimate goal is to create AI applications that are more dependable in complex and changing contexts, as the leveraged knowledge is more diverse, comprehensive, timely, and integrated, thus resulting in more precise decisions.
The lab will demonstrate the efficacy, reliability, and scalability of the trustworthy collaborative knowledge engineering solutions developed by the lab through pilots and Proofs-of-Concept (POCs) in collaboration with the industry partners in GENIUS (i.e., Biomax and DSM). In the long term, this can lead to the continuous integration of human intelligence into AI systems to increase the trustworthiness and dependability of knowledge engineering processes. The trustworthy collaborative knowledge engineering solutions can also result in improved performance of knowledge-driven applications (e.g., semantic search, question-answering) in commercial contexts and for the benefit of society at large.
The lab will innovate in i) the collaborative, agile, meaningful, and sustainable knowledge engineering between people and with AI systems, ii) the accurate and reliable integration, repair, and refinement of internal and external knowledge, and iii) the development of safe, resilient, and trustworthy knowledge-based applications. The fundamental scientific AI challenges addressed in this context by GENIUS (and corresponding to the five pillars in the lab’s setup for the PhD students) are: 1) collaborative knowledge synthesis, 2) integration and linking of distributed knowledge fragments, 3) coordinated knowledge repair, refinement and completion, 4) AI-supported text-based knowledge capture, and 5) human-AI ethics.
The joint work by the 5 PhD students and the accompanying researchers and domain experts allows researchers to co-produce methods and techniques for distinct yet highly interweaved knowledge engineering steps: collaborative knowledge synthesis, integration and linking of distributed knowledge fragments, coordinated knowledge repair, refinement and completion, knowledge capture, and human-AI ethics. The PhD projects together contribute to the development of a human-centered, trustworthy collaborative methodology across all knowledge engineering steps. These methods and techniques are developed, experimented with, and further matured in the contexts offered by Biomax and DSM where contextually sensitive and scientifically evidenced food and drug health claims can be used for decision support in the context of personalized nutrition and personalized medicine, where understanding the full provenance of consumables, including the who, what, where and how of manufacturing, distribution, and retail can be studied, and where the approach helps iterative refinement of domain ontologies to improve knowledge-driven applications. By making the scientific contributions available for research outside of the lab, other sciences and domains will be able to multiply the outputs from this lab.
