| Popis: |
Our society is currently divided into two knowledge ecosystems. Know4All, as a public decentralized initiative, shares knowledge between peers under demand, while the private ecosystem GoKnow centralizes knowledge management and data services. The community-driven efforts in developing resources for Know4All have lead to the construction of KnoMo, a system composed of an aggregate of resources that enables peer-to-peer data sharing. These modules have been developed for each knowledge domain, and are well-established and maintained by the community. This maintenance and curation process, however, is still carried out manually. We propose Evo-KnoMo, a framework that automates the evolution of the KnoMo Modules. It is based on usage metrics, meta-neuro-genetic algorithms, and explainability algorithms in order to automate the deprecation, search and implementation of new resources; while producing a readable report of the modifi cations with their explanation. We test this framework in a user study with the Healthcare and Transport domains with users and curators. Results show that Evo-KnoMo can evolve modules with a high accuracy rate and a considerable level satisfaction by the participants. We believe that this framework can help in the maintenance of KnoMo and thus, enhance the user experience for all users within Know4All. Disclaimer: This paper is a work of fiction, written in 2023 and describing research that may be carried out in 2043. For this reason, it includes citations to papers produced in the period 2024-2043, which have not been published (yet); all citations prior to 2024 refer instead to papers already in the literature. Any reference or resemblance to actual events or people or businesses, past present or future, is entirely coincidental and the product of the author’s imagination. |
