Computation of Asymmetrical Semantic Document Relations

A. Klahold, A. Holland, and M. Fathi (Germany)


Man-Machine Interfaces, Cognitive Processes, Knowledge Acquisition, Linguistics, Natural Language Processing, Software Tools


We present a new approach to create document recommendations for a Man-Machine Interface. Because of the determined asymmetrical semantic relations between documents the recommendations resemble a navigational system for the document corpus. The approach reduces the probability to get stuck in a group of documents semantically strongly related to each other. Instead with each new document selected there are mostly new documents recommended. That is because new recommendations are selected from the semantic viewpoint of last selected document. A special tool provides the visualization to explore these asymmetrical document relations. A study showed that the presented approach nearly equals the quality of the cognitive process of establishing semantic relations between documents by professional editors.

Important Links:

Go Back