Combining Bottom-Up and Top-Down Generation of Interactive Knowledge Maps for Enterprise Search

Endogenously Emergent Information Systems

“…The challenge in this situation is to make sense of the meaning (semantics) of all that heterogeneous information content. It is hard to believe that any a priori defined ontology could solve the issue, but it must take place more inductively in the spirit of emergent semantics [3,45]. For example, when a higher-level construct (e.g.…”

"Applying fuzzy ontologies to implement the social semantic web" by Edy Portmann, Patrick Kaltenrieder and Noémie Zurlinden with Martin Vesely as coordinator

“…Another possible approach is to combine top-down and bottom-up knowledge engineering, which has already been used to successfully create an enterprise search platform [Kaufmann et al 2014]. Top-down methods are able to project existing ontologies from the conceptual level to the document level to structure the data and explicitly represent the corporate context.…”

“…However, they cannot fully describe the semantics of concepts and their relationships, and they do not account for the organizational context. To overcome their shortcomings and to automatically generate a knowledge map, the two approaches can be combined by, for example, domain and enterprise ontologies [Kaufmann et al 2014]. …”

“…The enterprise ontology can be used as a basic structure graph that is gradually extended with information extracted bottom-up by assigning a degree of membership [Zadeh, 1965] to each concept predefined in this ontology. This provides the flexibility in the concept mapping that is needed for the integration of the two ontologies and allows for a simple ranking of search results [Kaufmann et al 2014].…”

“…FCMs [Kosko, 1986] can be useful for knowledge-based reasoning because conclusions can be drawn from the knowledge stored in the ontologies [Portmann & Kaltenrieder, 2015]. A Graphical User Interface (GUI) enables interactions between the user and the FCM, so that the user can browse it, search it by keyword, change it, and view the resources related to the concepts [Kaufmann et al 2014;Portmann, 2013]. Functions inspired by GrC [Lin, 1997;Zadeh, 1998] (e.g., zoom or drag-and-drop) support the efficient investigation of information and the automatic addition of new analytical content [Portmann, 2013].…”