HACID develops a novel hybrid collective
intelligence for decision support to
professionals facing complex open-ended
problems, promoting engagement, fairness and
A decision support system (HACID-DSS) is proposed that is based on structured domain knowledge, semi-automatically assembled in a domain knowledge graph from available data sources, such as scientific and gray literature.
Given a specific case within the addressed domain, a pool of experts is consulted to (i) extract supporting evidence and enrich it, generating a case knowledge graph (CKG) as a subset of the DKG, and (ii) provide one or more solutions to the problem.
Exploiting the CKG, the HACID-DSS gathers the expert advice in a collective solution that aggregates the individual opinions and expands them with machine-generated suggestions. In this way, HACID harnesses the wisdom of the crowd in open-ended problems, relying on a traceable process based on supporting evidence for better explainability.
A set of evaluation methods is proposed to deal with domains where ground truth is not available, demonstrating the suitability of the proposed approach in a wide range of application domains.
The goal of HACID is make the HACID-DSS deployable in diverse application sectors through a participatory AI approach, demonstrating the potential to readily reuse the concepts developed within the project across domains. Decision support for open-ended problems will be demonstrated in both medical diagnostics and climate services for urban adaptation.
The goal of HACID is to provide a participatory approach to evidence synthesis, to co-create the underlying knowledge representation and (semi-)automatically populate the knowledge base to obtain a large, navigable collection of evidence that can be exploited for reasoning and decision support. Experts are guided in the identification of relevant evidence in support of a specific case, resulting in the definition of structured knowledge that supports the hybrid collective intelligence in producing an optimal set of solutions.
The goal of HACID is to reason over the available knowledge to optimally aggregate advice from multiple experts, expanding it on the basis of the knowledge available to provide improved decision performance (e.g., accuracy) while maintaining explainability. HACID will determine context-specific approaches to tap into individual knowledge by eliciting confidence and justifications of proposed solutions, weighing individual expertise on the basis of the history of interactions and exploiting social influence by building suitable interaction networks among experts that can promote better decisions by overcoming individual and social biases.
HACID is an HORIZON Innovation Action, a collaborative project funded
under the Horizon Europe Programme, within
the topic "AI, Data and Robotics at work" (HORIZON-CL4-2021-DIGITAL-EMERGING-01-10).
It involves the following institutions: