Topic maps as a tool for global in - house knowledge management

Abstract

This thesis explores main avenues in the field of knowledge representation and resource classification and retrieval. As for knowledge representation, the solutions we have taken into consideration go from simple keywords to complicated domain representation structures, such as ontologies. We have looked both into the capacity of different techniques to picture the particular view some community has of a given domain, and into the ease of use and understanding of the associated representation language (graphics, text). For this reason (expressiveness), some techniques lacking of formal foundations or solid scientific basis (Mind Maps, Concept Maps) have been commented alongside with more strict ones. The Topic Map paradigm has attracted our attention because it fulfills formal and cognitive constraints that make a powerful tool for knowledge representation and semantic resource identification out of it.

After comparing its benefits with other approaches of the problem, we have implemented

a full web tool to manage this type of structures and bind documentation to them. Considering the weight of XML data flow in nowadays culture, we have implemented the extraction of information from this type of format, which allows Topic Maps to contain not only full body resources but also document chunks.