Aspects of Text Mining From Computational Semiotics to Systemic Functional Hypertexts

Abstract

The significance of natural language texts as the prime information structure for the management and dissemination of knowledge in organisations is still increasing. Making relevant documents available depending on varying tasks in different contexts is of primary importance for any efficient task completion. Implementing this demand requires the content based processing of texts, which enables to reconstruct or, if necessary, to explore the relationship of task, context and document. Text mining is a technology that is suitable for solving problems of this kind. In the following, semiotic aspects of text mining are investigated. Based on the primary object of text mining - natural language lexis - the specific complexity of this class of signs is outlined and requirements for the implementation of text mining procedures are derived. This is done with reference to text linkage introduced as a special task in text mining. Text linkage refers to the exploration of implicit, content based relations of texts (and their annotation as typed links in corpora possibly organised as hypertexts). In this context, the term systemic functional hypertext is introduced, which distinguishes genre and register layers for the management of links in a poly-level hypertext system.