Knowledge Space Vectors in Multi-Dimensional Classifications

Activiteit: Toespraak of presentatieAcademisch


Artificial intelligence produces large quantities of morsels of knowledge that require multi-dimensional classification. Library and information science has a long history of representations of multidimensional classification of such morsels. Spatial metaphors were used to classify ‘things’ (Richardson), ‘facts’ (Otlet) and ‘concepts’ (Beghtol) successively in universes of knowledge and of concepts. The gradual change from a philosophical perception of classification toward a more pragmatic one in support of information retrieval also changed the representation of multi-dimensional classifications. Imaginary space travel in planetary systems of knowledge gradually made place for queries for information following vectors in rigid geometrical knowledge spaces such as cubes. By comparing geometrical knowledge spaces of Otlet, Dahlberg and Meincke & Atherton, this paper discusses the transition in the conceptualization of concepts (from more static toward evolutionary and interactive) and the implications hereof for representation of multidimensional classification. For this comparison we will use both contemporary visualizations and digital reconstructions of multidimensional knowledge representations to analyze in particular problems in bringing in multiple dimensions of concepts in the transition from two- to multidimensional knowledge spaces and the retrieval hereof. In particular we will zoom in on discussions between philosophical and computational approaches (Dahlberg and Bowker) in the multidimensional classifications and claim that these are still relevant for getting a grip on graphs as knowledge representations.
Periode21 jun. 2019
Mate van erkenningInternationaal